The InOrbit team recently had a chance to showcase Mission Control, our cloud-based robot management platform, at RoboBusiness. This event has become one of the most important in the industry, attracting the most innovative robotics companies as well as a business audience hungry for technology to help tackle some of the hardest problems across different industries.
We had a really busy week, including co-organizing the first face-to-face meeting of the Robot Operations Working Group, presenting on stage in front of +100 people at Pitchfire, and moderating a panel on Best Practices for Robotics Operations at Scale, with great panelists from Brain Corp, Qualcomm and Service Robotics & Technologies.
We also had a lot of fun preparing our demo for the show. Check out the video below.
Since the work we do is somewhat abstract (see Bridging the Autonomy Gap), we wanted to make it more concrete, so we built a mini-grocery store/warehouse we called “Brickmart” … out of Legos, of course. Our engineering team heavily modified a TurtleBot3 running ROS, including adding support for multiple HD cameras with hardware video encoding and upgrading to a Raspberry Pi 4. As if this wasn’t enough, our team had InOrbit running on the recently released Qualcomm® Robotics RB3 platform.
Our unique and colorful display, with two autonomous robots, generated a lot of interest. People stopped for the cool demo, but stayed for the business value: we were demonstrating how InOrbit’s real-time analytics and incident management can improve robot fleet efficiency. With a personalized experience for business executives, robot operations managers and individual operators (we like to call them roboteers), InOrbit allows monitoring overall fleet health and resolving issues remotely.
We showed how, in response to an autonomy exception such as a false positive from the obstacle detection algorithm due to a reflection or a mis-localization, InOrbit allows a roboteer to get instantaneous situational awareness, re-localize the robot in seconds, decide to resume operations with a single click or take more active steps, such as setting waypoints for the robot to navigate around a (real or imagined) obstacle.
As a way to show off our engineering chops, and to make it even more fun for attendees, we also demonstrated low-latency video streaming and real-time teleoperation through the cloud with a gaming steering wheel. Whle this may not be as much fun as the latest Forza racing video game, it is pretty cool to show how people and autonomy can be combined to achieve results that neither could on their own.
If you’d like to get a personalized demo of InOrbit, sign up at inorbit.ai/promo.
Our goal is to put every robot in orbit around the cloud to help accelerate the adoption of robotics across industries. Today we are one important step closer to that goal.
At RoboBusiness, the premier commercial robotics trade show for business executives, we announced that the Qualcomm® Robotics RB3 development kit will have pre-integrated support for InOrbit’s cloud platform. This advanced development kit is based on the powerful Qualcomm SDA845 SoC. This allows robotics developers to create autonomous robots for the most challenging applications.
The Qualcomm Robotics RB3 development kit, available for purchase from Thundercomm is based on the Qualcomm® Robotics RB3 platform, and supports the development of smart, power-efficient and cost-effective robots by combining high-performance heterogeneous computing, Qualcomm® Artificial Intelligence (AI) Engine for on-device machine learning, computer vision, vault-like security, multimedia and Wi-Fi and cellular connectivity.
“Having InOrbit as part of our robotics ecosystem is key to helping accelerate the adoption of robotics at scale,” said Dev Singh, director of business development and head of robotics, drones, and intelligent machines, Qualcomm Technologies, Inc. “The combination of Qualcomm SoCs on the robots and InOrbit in the cloud allows companies to zoom in on application-specific needs.”
As autonomous service robots are deployed at scale, the need for connectivity to the cloud is becoming a critical factor for scaling operations. Qualcomm is best known as the leader in smartphone chipsets and cellular communication, and with RB3 the company has created a great platform for rich robotics applications.
As 5G connectivity is rolled out, autonomous robots based on RB3 will be able to connect to the InOrbit cloud and benefit from low-latency, high-bandwidth connections. This will help advance our vision of humans, robots and AI in the cloud driving advancements in productivity.
In Part 1 of this article, we discussed in some detail some of the limitations of AI and autonomous systems. The key takeaway is that autonomy is relative, and there continues to be a need for human interaction and direction.
At InOrbit, we are harnessing the power of the cloud and the edge to bring automation and efficiency to the operation of distributed robot fleets. It consists of four O’s, which of course we think of as concentric orbits.
It all starts with Observability. Robots can easily generate terabytes of data per day. The needs for access to the data also vary significantly, from robot health monitoring to teleoperation. Making the necessary data available and actionable requires careful balancing of resources such as computation, storage and network. We address this with what we call adaptive diagnostics, and it’s a key part of our very special sauce. The data captured in the cloud includes time series data, logs, video and costmaps, to name a few.
Once the detailed operational data is available, we can build on it to drive Optimization of the actual deployments. This can range from categorizing problem areas to identifying the root cause of an issue. We also help drive improvements in some key business metrics, such as the Total Cost of Ownership/Operation (TCO) for our customers.
The next orbit is Orchestration. We believe that real world problems will be tackled by a variety of task-specific machines, vs. general purpose robots. A large construction site may need a different mix (ranging from heavy autonomous earth movers to inspection robots that can crawl into confined spaces) than an e-commerce distribution center (self-driving forklifts, de-palletizers, inventory management drones). Operating all these machines together, including managing robot-to-robot (R2R) and human-to-robot (H2R) coordination, requires a new layer of cloud software. InOrbit can provide a single pane of glass to manage all this complexity.
Building on all of the above is Operation. Robots in the field require different levels of monitoring and interventions. In order to enable this at scale, InOrbit provides managed services for level 1 and level 2 support, augmented through our AI and data platform. This allows robotics companies to focus on improving the core software for their specific application.
This week we will be demonstrating our solution at RoboBusiness, the premier trade show for robotics, in Santa Clara, CA. We’ll also be sharing more in future posts about specific steps we are taking to boost adoption of robotics across industries.
Imagine that you need to fly to a distant location, a busy airport you’ve never visited. You are the nervous kind, so you want to make sure it will be safe. You talk to the airplane manufacturer, and walk away with confidence that the machine has been well designed. You verify the maintenance schedule and are satisfied that it is thoroughly tested.
Then you talk to the management at the airport you’re flying into, and they tell you that they cobbled together their air traffic control system with some old PCs that nobody was using and installed some software for a taxi dispatch. Would you get on that plane?
Sadly, this is often the case in robotics. Since before we started InOrbit, and throughout the last 2 years, we have engaged with over 75 robotics companies and hundreds of people in the robotics space. We have talked to the C-suite setting strategy and to front line operators doing triage, to robotics Ph.D.’s and to self-identified robot baby-sitters. Across all these conversations, we had an overarching question: how do you manage robots after they leave the lab?
To summarize the response from the majority: Not well. This has motivated us to continue working on a way to help these companies accelerate. We also kept digging deeper, spending quite some time understanding why most robotics companies find themselves in this situation. For the most part, what we’ve learned is that, while effective robot management in the field is absolutely critical, for most companies it is not their core expertise.
The journey that most robotics companies follow is remarkably similar. For many of them, it starts with “robots are cool, let’s build one”; what the robot is for is a secondary consideration. Thanks to great advancements in tools and components over the last 5 years, the time and cost of building an initial prototype in the lab has dropped dramatically, so this story is repeated hundreds of times each year around the world as new startups are founded.
After getting their first robot working in the lab/garage/accelerator/maker space, it’s time for the initial deployment. It may be a field trial in a particular location, a proof of concept or a pilot with a corporate customer. The number of robots goes from a handful in the lab to a dozen or two in the field. Now some of the practices and tools that were meant for development in the lab start to show their limitations, but they can still largely brute force it: the company has raised some funding and has more engineers than robots.
The next stage is scaling, from dozens to hundreds and then thousands. This is where many companies crash land. Manufacturing can be outsourced, most components are off the shelf, and the robot software has been refined during the pilot. However, their operations are not ready for scaling. Their robot : human ratio is upside down, they can’t hire qualified people fast enough, their tools are clunky, inefficient and brittle.
Until recently, robotics companies didn’t have a choice but to build their own. For most of them, this was at best a distraction or side project; all the cool kids were building robots. A few select companies have recognized the need of investing heavily to develop these tools and best practices in-house. However, their learning is severely limited by the myopia of only looking at their own robots.
This is now changing. InOrbit and a few companies following in our footsteps are bringing scalable tools to help robotics companies and operators better utilize their robots in the field. In our case, our main focus is on RobOps, which combines the best practices and tools of DevOps and AI to bridge the autonomy gap and orchestrate the operations at scale of heterogeneous robot fleets.
There’s no longer an excuse to “cobble together” some disjoint tools. Companies facing the build vs. buy decision can now saw “Yes, and” by cherry-picking specific elements of the InOrbit platform, easily customizing/extending the functionality to meet their needs, combining it with their existing home-grown tools or adopting it as a ready-made cloud solution to run a complete ROC (Robot Operations Center).
Robots are everywhere. They can be found in hospitals and hotels. In farms and construction sites. Brick and mortar retailers and e-commerce distribution centers. In the air, in the sea, on the ground and even underground, as we saw recently in the DARPA challenge.
But what is a robot? There have been plenty of philosophical discussions on this, and probably no shortage of flame wars. We like this definition from IEEE:
A robot is an autonomous machine capable of sensing its environment, carrying out computations to make decisions, and performing actions in the real world.
So right there in the definition is the A-word: autonomy. Since nomos is Greek for “law”, something autonomous makes its own laws. Pretty cool, right?
However, autonomy is relative. We’re not just talking about being constrained by the laws of physics, but by the limits of AI, sensing technology, computing power, servos, etc. In essence, the guts and brains of robots can only go so far with current technology.
Surely given all the brilliant minds that are working on this problem, we should have complete autonomy in a couple of years, right? We don’t think so. In fact, complete autonomy may never be possible.
In the self-driving car space, the SAE (formerly the Society of Automotive Engineers) has defined the six levels of driving automation, ranging from no automation (level 0) to full self-driving automation (level 5) requiring no driver intervention – and perhaps no human passengers either.
Getting to level 5 will take a while. Jim Hackett, CEO of Ford Motor Co. acknowledged as much recently: “We overestimated the arrival of autonomous vehicles.” John Krafcik, CEO of Alphabet’s Waymo, made a stronger statement: “Autonomy will always have some constraints.” More to our point, even level 5 does not mean complete autonomy in terms of being self-directed.
Although these challenges were discussed in the context of self-driving cars, they apply more broadly to (other) robots. Small digression: a self driving car is really a ground-based robot with a hole in the middle for carrying cargo called people, and even that is not a hard requirement – think trucks without a human cabin.
For instance, ground-based autonomous mobile robots are deployed in various environments, usually for material movement or data collection. The applications are diverse, from campus security to inventory management at a retail store or goods-to-person delivery in a warehouse. These robots typically use a combination of odometry, RGB cameras, depth cameras and LIDAR to sense their environment.
While these sensors, and the computer vision algorithms that are used to make sense of the signals coming from the sensors for navigation purposes, have advanced tremendously in the last decade, they still have limitations. Sometimes, a reflection off a particularly clean floor is perceived as an obstacle. Other times, a small slip of the wheels may result in “mislocalization”: the robot isn’t sure where it is in the map, and is therefore stuck without some assistance.
At InOrbit, we call this the autonomy gap. We’re on a mission to bridge this gap by creating DevOps and AI tools as well as developing best practices to drive human-in-the-loop robot operations at global scale.
In Part 2 of this article, we’ll share more about InOrbit’s plans to help the robotics industry accelerate the adoption at scale of autonomous robots by bridging the autonomy gap.
50 years ago this month, the first human set foot on the moon (unless you choose to believe it never happened.) At InOrbit, we work on scaling autonomous solutions, not spaceships, but this milestone got us thinking about what the next 50 years may have in store for us, and the impact of automation.
As I’ve shared here before, we do 3-4 big product pushes per year and give them a codename based on some of the greatest inventors and scientists. It seems appropriate that for our next one, which will be starting soon, we would pick Margaret Hamilton.
While the picture of Neil Armstrong’s first footprint on the moon has been repeated over and over, it’s not as often that we hear about all the people who made that possible. Hamilton led the software development efforts for Apollo 11; she was the first person to use the term “software engineering”. She was awarded the Medal of Freedom in 2016. If you’d like to learn more about Margaret Hamilton, there’s a great profile on Makers.com.
Besides being a pioneer in space exploration, something that obviously resonates with us at InOrbit, her work is still relevant to our current efforts. Hamilton designed Apollo’s software to be capable of dealing with unknown problems and flexible enough to interrupt one task to take on a more important one. One of our goals at InOrbit is to make robots safer and more resilient by identifying, resolving and anticipating problems that occur in the field by harnessing the power human problem-solving abilities and AI in the cloud.
More generally, the code that Hamilton’s team designed had to run in a resource-constrained environment: even if today’s robots have millions of times more storage and processing power than the Apollo 11’s two 70-pound computers, they are still limited compared to the processing that’s possible in the cloud and are often connected through limited bandwidth.
Looking at the next 50 years of space exploration, we believe that robots will play an integral part. Beyond the Mars Rovers, robots will be able to inspect and repair spaceships, under commands triggered by humans and/or AI. Moreover, before any settlement in space or on another planet, robots will likely be sent ahead of humans in missions to set up a living environment.
Today autonomous robots are taking some big steps – and also rolling, flying, hopping, swimming, balancing, slithering and any conceivable form of transportation. While there are many immediate applications of robots right here on Earth, which will transform our lifestyles over the same 50 years, thinking about robots and space exploration gets us excited at InOrbit. After all, our goal is to put every robot “in orbit” and our first product is called Mission Control.
Over the last year, we have been talking with robotics companies and operators in Silicon Valley and around the world. We have learned a few things about the challenges of operating robots at scale. This has mostly confirmed that, compared to the great advances in software and hardware components that are helping robots enter every industry, robot operations and infrastructure are still immature.
We have found that most robotics companies are using tools designed for developers in the lab to operate robots in the field. This is often an after-thought or a distraction from the “cool work” of building robots. In the words of the CEO of a successful robotics company with hundreds of robots in the field: “our engineers have cobbled together a bunch of tools.”
Is this how companies should be managing costly machines that weigh hundreds or thousands of pounds and move around on their own amongst people? It is more than a little ironic that, for an industry that is bringing about massive changes through automation, many of its own operational practices are extremely manual, repetitive and error-prone.
Part of this may be explained away as a matter of industry maturity. The number of companies with thousands of autonomous service robots in operation is still relatively small. However, the journey that robotics companies follow as they scale is markedly similar, regardless of their specific application or industry. Robotics startups that may not yet be feeling the pain of trying to run or support hundreds of robots in far-flung locations would be well served by learning from their more advanced brethren.
Below is a list of some of the most egregious issues we’ve found in robot operations. Not every company is incurring all of them or to the same degree, and some of the companies we are working with at InOrbit are already far ahead of the curve, but as a whole this paints a pretty grim picture – or rather, there’s a lot of room for improvement.
Poor security practices
Many robots share the same access password or have generic management accounts as a way to reduce the management complexity, at the expense of reduced security
Reliance on SSH
Creating an SSH connection and running tools over a point-to-point connection is a catch-all solution in the absence of proper tools designed robot operations in the field
Inadequate incident response
When an issue occurs, it can take a long time for the problem to propagate and tracking of the resolution is often incomplete, making it hard to manage or get to the root cause
Lack of real-time notifications
Related to the previous point, many systems are able to perform limited logging but are missing a real-time mechanism to detect and respond to issues as they occur
Frequent navigation exceptions
Deployment in the real world is messy, resulting in frequent issues that result in the robot leaving autonomous mode, reducing operational efficiency and asset utilization.\
Running scripts and commands manually
Robot configuration and management is often done by running sequences of commands or scripts manually, resulting in “snowflake” robots that are impossible to manage at scale
Networking issues including VPN traversal
It is common for roboticists to continue using in the field tools that were meant for the lab where they run on the same local network, but operating in customer-controlled networks
So what can be done about it? Plenty.
In terms of tooling, the InOrbit platform offers solutions to most of the issues above. Given our robot-agnostic approach and ease of platform extensibility, InOrbit can be deployed quickly for any size robot fleet, including heterogeneous fleets with different types of robots for different tasks. If you want to dig deeper, you can try InOrbit for free, just go to inorbit.ai/quickstart.
However, our mission goes beyond that: we want to help robotics companies get to operations at global scale. To get there, the industry needs to build operational best practices. We are helping create a movement, similar to the adoption of DevOps practices and tools.
To that end, we have founded a Robot Operations Working Group. We are starting small, with a few experienced leaders and passionate robotics practitioners. Our first virtual Meetup will be later this week. If you are interested in joining this “cross-industry group aimed at developing and sharing best practices and advancing the state of the art for the operation at scale of autonomous robots”, you can learn more at ROWG.
Another day, another Demo Day. Last week we had the privilege of being among a select group of startups to show the latest and greatest in the IoT world, ranging from photonics to wireless power transmission, at the latest Plug and Play Demo Day for Internet of Things. This was a big event, with many corporate partners and investors in attendance.
Our focus at InOrbit is on driving the new field of RobOps – DevOps for Robotics. Our mission is to accelerate the adoption of robotics across all industries and applications, from food production to retail. We are doing that by bringing to the robotics community many of the best practices and tools that allowed the cloud to scale. This results in better productivity for roboticists and lower time to resolution for ops teams.
We are often asked why we focus on robotics versus creating a generic IoT platform. After all robots are just things, right? (At this point, I’m tempted to go with “robots are people, too” but we’ll leave the philosophical and ethical discussions around personhood and robot rights for another post.) Since we were just in the Internet of Things batch, I thought I’d cover our thinking on the unique needs and challenges of robotics that set it aside from the more generic IoT.
In typical IoT scenarios, each node is a standalone sensor reporting data that is then aggregated in the cloud to make decisions. In the case of autonomous robots, each robot is typically includes its own _self-contained constellation _of sensors (RGB, depth stereo cameras, lidar, distance/proximity/bump sensors, GPS, etc.) and actuators (servos, brushless motors, grippers, etc.) By self-contained, we mean that the data generated by these sensors is primarily used locally by the robot’s on-board processor to build a real-time model of the world and make decisions on it, in turn sending real-time commands to the actuators.
Many modern robots use the Robot Operating System (ROS) as a way to manage this massive data flow. Technically more middleware than operating system, ROS includes capabilities for hardware abstraction and message passing to integrate these various data sources.
For instance, a typical ground-based autonomous robot may have four HD cameras, two 3D depth sensors, and one or more lidars. If we include input and output from robot processes doing AI (computer vision, path planning), this adds up to 500Gb/h.
To make matters worse, many robots run on networks outside of the control of their developers and/or operator. It is not uncommon for service robots to be deployed on customer’s premises, outdoors, in public areas or constructions sites. In addition, most robots have some level of mobility, whether moving from one location to another (AMRs, drones, autonomous forklifts, etc.) or manipulating physical objects (robot arms with 6 degrees of freedom, etc.)
All this adds up to a number of unique challenges. At InOrbit, we have built a platform specifically geared towards the kinds of data sources and operating environment the autonomous robots typically handle so that we can properly complement their on-board capabilities with cloud-based functionality.
Partly as a corollary of the previous point, modern autonomous robots can generate several orders of magnitude more data than, say, a smart thermostat. This requires a different approach to enable remote management.
One of the key features of InOrbit is the ability to provide robot performance monitoring through real-time analytics. However, a naive approach of sending all the data to the cloud just wouldn’t work. Robots in the field typically operate in constrained network environments: spotty WiFi coverage in warehouses or retail stores, limited 4G bandwidth, etc.
Part of InOrbit’s secret sauce (aka patented technology) is what we call adaptive diagnostics, which allows us to process massive amounts of data at the edge and make only the most critical information available in the cloud as needed.
Robots can sometimes operate on their own, relying purely on their on-board sensors. However, real world deployments will typically have a variety of sensors and often multiple robots, most likely from different vendors.
Since InOrbit has already tackled the much harder problem of data analytics and remote management for robots, it is relatively easy to integrate other devices. For example, a robot may operate in a warehouse where there are fixed cameras, RFID readers, proximity sensors at loading docks, etc. The same warehouse may use heavier equipment such as autonomous forklifts for moving pallets, fixed arms for de-palletizing and goods-to-person systems.
InOrbit can help orchestrate all of this through the cloud, creating a “single pane of glass” for all the robot and sensor data. In our warehouse example, operators can have a full operational view that provides real time data as well as the ability to take coordinated actions when interventions are needed.
To wrap up, on the question of Robotics or IoT, we’ll take a page from improv and go with “Yes, and.” Robots can be thought of as either a subset or a superset of IoT devices. InOrbit is tackling the specific needs of managing autonomous robots that generic IoT data platforms can’t handle. With this more advanced capability, it is easier to integrate other data sources from simpler IoT devices and offer a single dashboard across devices regardless of their level of complexity.
If you want to dig deeper, you can try InOrbit for free. Go to inorbit.ai/quickstart to get started.
In the last couple of years, it seems that robots have jumped from science fiction to reality and are on their way to becoming commonplace. This process continues to accelerate, driven by macro-economic trends such as labor shortage and advancements in technology.
But as I like to say, it usually takes years of preparation to become an overnight sensation. In the case of robotics, it has taken decades of contributions from people around the world, from scientists to entrepreneurs and investors.
The InOrbit team got together for an all hands a few weeks ago to do the groundwork for our next major milestone. We’ve adopted the practice of giving each of these big stakes in the ground a codename based on some of the greatest inventors and scientists. We are just now wrapping up Rita, which was named after Rita Levi-Montalcini.
For the next one, we are honoring Nils Nilsson. A computer scientist who was one of the founding fathers of robotics, artificial intelligence and machine learning. He passed away in April, 2019.
Nilsson worked at Stanford Research Institute (now SRI International) for over 20 years starting in the early ‘60s at the Artificial Intelligence Center. He then joined Stanford as chair of the Department of Computer Science. Amongst his many contributions was “SHAKEY”, one of the first autonomous mobile robots, as well as several algorithms that are the predecessors of current robotics algorithms.
As it turns out, I’ll be at SRI today as part of The Rise of the Robots, a Japan and US Robotics conference organized by Silicon Valley Robotics. I will get to pitch InOrbit to an international audience of roboticists and investors.
I have recently had the honor of being introduced to another visionary and unsung hero of the robotics revolution. In this case, it was Buck Ward who 30 years ago envisioned using robots to clean floors, relieving workers from the repetitive tasks and offering improved reliability to his customers.
I was so inspired by connecting with one of the early pioneers in the space of autonomous robots and learning about his journey. Reality is only now catching up with his vision. I was struck by this article from the LA Times 22 years ago. The discussion and some of the quotes are as relevant today as they were then.
“I recognized a long time ago that the cleaning industry is the same as it’s been for the last 1,000 years,” he said. “Most of the work is done with a rag and some elbow grease. . . . The entire computer industry has pretty much bypassed the cleaning industry.”
However, it is only now that the technology has matured and the cost has come down sufficiently to make this not just economically viable but practically unavoidable. Cyberclean Systems, the company that Buck started all those decades ago, is now growing rapidly in a RaaS (robots as a service) model powered by autonomous floor-cleaning robots from different vendors.
Lastly, at a personal level, my wife’s father passed away yesterday. Although he was not directly involved in robotics, he spent his whole life doing what he loved best: decoding the inner workings of the human brain. As a scientist, he published too many books to list here, trained many leading international researchers and pioneered computer-based language analysis techniques.
It is at times like this that we re-discover that we truly stand on the shoulders of giants. To help advance the state of the art in robotics, we are forming the Robot Operations Working Group or ROWG. The purpose of this cross-industry group will be sharing and developing best practices for the operation at scale of autonomous robots. If you’re interested, please find additional details and sign up at www.inorbit.ai/rowg.
Few things are more devastating for a startup than to be working on the wrong problem. That’s why we spent time early on, before we had even formed our company or written the first line of code, to make sure we were working on a real problem that has a significant impact for a large number of people. We met with dozens of people across the robotics landscape to learn, asked lots of open-ended questions and listened carefully, honing our pattern-recognition.
To be honest, we actually cheated a bit. Or a lot, actually. We started this process with an unfair advantage: my co-founder Julian’s experience over the previous +7 years building software for robotics companies, often times building roughly the same solution from scratch to solve the same problems over and over. This gave us a running start, knowing in depth the strengths and pain points of influential companies, including Willow Garage, Clearpath Robotics, iRobot and Savioke to name a few.
During the better part of 2018, we were heads down on solution validation, working very closely with a small set of committed and very patient early adopters, who helped us refine our product with their invaluable feedback, suggestions, bug reports, and general fandom. We are grateful to Fellow Robots, Iron Ox, Rover Robotics and many others. Along the way, we also kept talking to people in the robotics community and iterating, redoubling our efforts in areas that were promising and leaving behind some ideas that did not seem to catch on. We learned some of what worked and what didn’t, and this got us to the end of the year with our very first (modest) revenue.
But you can never know too much about your customers … at least in the B2B world, where this is not as creepy as it sounds for consumers. I like to say that a startup is a learning machine – not to be confused with machine learning (hint: we do that, too.) So you have to be constantly learning new things. That’s why this past week was a big deal for us, as it marked the confluence of learning efforts that we started at the beginning of the year.
We applied to Plug and Play’s Supply Chain and Logistics accelerator before the end of the year. Plug and Play is the ultimate innovation platform, connecting the best technology startups and the world’s largest corporations. In fact, we were one of 1,000 startups that applied to this program. The list got whittled down to 22 through the votes of leading corporates, including those shown below. Then we did it all over again and were selected for the IoT batch.
We’ve been parallel-tracking in these two batches and have met with over 25 of the leading companies to learn about their plans for automation. Last week was the Plug and Play Spring Summit, a demo day where we got to pitch in front of hundreds of attendees (video), most of which were representing corporates across several industries. Through this process, we were able to validate our go-to-market hypotheses and we learned last week that we have been approved for a proof of concept (PoC) by one of the corporates in the program.
Speaking of PoCs, last week we also wrapped up another collaboration with one of the leaders in the robotics space. This company has hundreds of robots in the field right now and will have thousands by year end. Since they already have a mature infrastructure, our focus was to identify new capabilities and use cases that could be enabled through the InOrbit platform, including real-time analytics and remote interventions.
We concluded with a demo that was attended by over 20 people across the organization, including senior executives who provided invaluable feedback and shared ideas for how this could potentially be used to improve internal functions, from sales to product development. When we set out to build InOrbit, we knew that we had hit on some common needs across the industry and also expected that they would manifest in slightly different ways at each company, so we built flexibility and extensibility into our platform. One of the best feelings in tech is when people take something you’ve built and use it in ways you had never anticipated.
Before the week was over, we were able to squeeze in a meeting with one of the largest technology providers in the world to explore how our technologies can complement each other to help drive the adoption at scale of autonomous robots. We expect to be doing more in the coming months to help connect the various technology pieces and operational best practices, so please stay tuned. It takes a village (or an ecosystem) to tackle some of humanity’s most vexing problems by making it possible for humans, robots and AI in the cloud to work together.
One of our driving strategies at InOrbit has been “learning at scale”. Although we still have much to learn, as a company and as an industry, this past week was the culmination of many months of accelerated learning. Thank you to everyone who made it possible.
You can sign up for InOrbit for free by visiting our Mission Control center.
It seems that every week there’s news about a new amazing application of robotics, from e-commerce, to growing food or building houses. While for many people this may seem to be something for the distant future, it is happening now and will continue to scale rapidly over the next couple of years.
This rapid acceleration results from a convergence of factors, including advances in sensors, deep learning and computer vision algorithms, chipsets that can handle AI tasks and software standards such as the Robot Operating System (ROS).
This has many parallels with the development of the mobile ecosystem a few years ago. For decades, phones were dumb, like the industrial robots that build most cars today. Then, advances in hardware and software made it possible to create a smartphone with incredible flexibility, which led to an explosion in the number of applications.
We’re now seeing the advent of smart robots, including autonomous mobile robots, self driving cars and trucks; autonomous drones and unmanned aerial vehicles; special-purpose robots for pipe inspection; and many others. In a way, this feels like the months before the launch of the original iphone.
Taking an even broader look, the widespread adoption of new technologies in the past have had profound impact on productivity, which in turn have improved quality of life for most people. The current acceleration in automation is generating significant anxiety around the impact it is likely to have on employment. Without a doubt, as automation in general and robotics specifically are deployed at scale, change will inevitably happen in employment patterns.
However, much of the discussion to date seems to assume a fixed demand. History shows that as efficiency increases and prices drop driven by technology, demand grows to absorb much of the increased capacity, thus creating opportunities for everyone.
The best science fiction writers allow us to imagine a new world while helping us think critically about our current experience or conditions.
In the sci fi novel Roboteer, Alex Lamb introduces the main character Will Konu-Monet, who was “modified at birth to interface with the machines used to assist and maintain the infrastructure of his home world.”
The word roboteer isn’t new, but the notion of a human operator using technology to become a sort of robot wrangler seems compelling and necessary. Some of the most ambitious undertakings for humanity, from dealing with climate change, to large scale oceanic projects, to off-world settlements, will require a level of automation that doesn’t exist today, but is tantalizingly close.
Beyond sci-fi, there’s much discussion currently about levels of autonomy, specifically around self-driving cars. These levels of autonomy range from basic (driver) assistance to the elusive Level 5 offering full automation. However, even at Level 5, humans are required to provide direction: determining destination, when to leave or perhaps choosing the scenic route or a stop at a fast food place along the way.
Generalizing to other automation systems, research is currently underway on Human-Autonomy Teaming, based on the idea of humans interacting with increasingly autonomous systems as team members, rather than simply using automation as a tool. In the near future many of us will have robotic co-workers, and it will be as common as using a smartphone or a power tool.
All this will require careful orchestration of automation solutions. At InOrbit, we envision a world where humans, robots and AI work together to drive radical improvements in productivity and empower people to reach new heights.
As the author William Gibson wrote, the future is already here, it’s just unevenly distributed. We are already seeing some of the elements outlined above coming together.
Without going to the extreme of genetic modifications and hardware implants, we are already seeing companies setting up Robot Operations Centers (ROCs). InOrbit provides a cloud platform that enables companies running distributed fleets of robots to efficiently develop, deploy and operate automation systems at global scale. Or as I like to tell my team: we ROC!
Using InOrbit, operators – or, dare I say, roboteers – can monitor the health of their fleet in real time, get notified when there’s an issue, and resolve problems remotely. Using a mobile device or a browser, they can send the robot on a mission or even take full control and navigate the robot remotely and safely. Moreover, getting real-time information from distributed robots makes it possible to identify problems before they occur, by using our AI capabilities.
You can sign up for InOrbit for free by visiting our Mission Control center.
At InOrbit we are driven to advance the positive impacts of robotics in the world. Our vision is that humans, robots and AI working together can drive radical improvements in productivity and solve some of humanity’s biggest challenges. We believe that we can be a catalyst to innovation and allow robotics companies to focus on their special sauce.
Our product team has embraced the customer development model, which means we are constantly learning from and validating hypotheses with our customers. We are inspired by our customers, who are solving exceedingly hard problems in industries as varied as retail, logistics, construction, maintenance, security, healthcare and agriculture to name a few.
To get our team aligned and pointed in the same direction, every few months we get together and agree on the highest priorities. We are very much building the rocket even as we accelerate towards the stratosphere and beyond, so our goal is to identify the biggest impact items and then give the teams room to iterate towards the best solution.
Last year we started using codenames for these multi-month milestones and taking turns at picking these names to help our inspiration. For the current one, we picked “Rita” after scientist Rita Levi-Montalcini. It seems appropriate that on the International Women’s Day we talk about her life story and scientific contributions.
Rita Levi-Montalcini won the Nobel prize in 1986 in recognition for her work on nerve growth factor, which promotes the growth and maintenance of the nervous system. She was a scientific entrepreneur and the first Nobel laureate to reach 100 years of age; she died on December 30, 2012.
During the 1930s, shortly after her graduation from medical school in Turin, Jews were barred from academic and professional activities. “I then decided to build a small research unit at home and installed it in my bedroom,” she wrote in her biographical notes. Later on, after immigrating to the US, she created a lab in Rome and split her time between the two locations. Her work centered on neurons and the nervous system, specifically nerve growth factors. She founded the European Brain Research Institute in 2002.
In this age of Artificial Intelligence, in which we are complementing biological neurons with artificial ones, it seems appropriate to recognize the impact of pioneers in neuroscience. Rita Levi-Montalcini fought through deeply entrenched sexism and anti-Semitism to do the thing she loved most. At InOrbit we are similarly driven and building a company that we hope will last over 100 years, helping extend human intelligence to the stars.
InOrbit has been selected by the largest companies in Logistics and Supply Chain as one of the leading technologies to help drive automation at scale. From an initial set of 500 startups, 22 have been chosen by participating enterprises to be part of the Supply Chain Innovation program at Plug and Play.
This is a unique opportunity to engage with the leaders in logistics who are counting on automation to help them scale. Several macro-economic trends are impacting the movement of goods at global and local scale. Yutaka Nagao, CEO of Yamato Transport Japan’s largest package delivery company, said in an interview: “I expect the labor shortage to continue … We need to consider labor-saving measures.” (Source: Reuters) The company has announced it is developing unmanned flying delivery systems in collaboration with Bell Helicopter.
AI and robotics are changing the face of logistics and supply chain through innovations in inventory management, material movement, warehousing, maintenance and last mile-delivery, to name a few. “The demand for robots and the supply of advanced robotic solutions for the optimization of logistics processes, combined with labor shortages, have created a tipping point that could lead to widespread adoption of robots in warehouses and logistics operations to assist and displace human workers.” Shipments of warehousing and logistics robots are expected to grow rapidly over the next 5 years from 194,000 units in 2018 to 938,000 units annually by 2022 (source Tractica.)
InOrbit helps companies improve the efficiency of their robot fleet. Our cloud-based automation management platform enables enterprises to develop, deploy and operate smart machines at global scale. With over a dozen customers since our launch in 4Q2018, our analytics and AI system processes over 100 TB per day … and growing quickly.
The leading suppliers of logistics solutions are already deploying dozens of automation solutions. InOrbit allows them to control their growing fleets of heterogeneous robots from multiple vendors with a single analytics and operational platform for a complete view of their automation processes.
Through predictive analytics, ML-based actionable insights and AI algorithms, InOrbit orchestrates siloed automation solutions into an integrated system. As adopters go from proof of concept to pilots to at-scale deployment, we can put all of their smart machines in orbit and manage them through our Mission Control software.
The annual convention of leaders in robotics took place September 25-27 in Santa Clara, CA. RoboBusiness is the prime conference on the business of robotics, with world-class speakers and top exhibitors from around the world.
At InOrbit we chose RoboBusiness for our commercial launch. The interest far exceeded our expectations: the traffic at our booth was virtually non-stop, often with multiple conversations going on with members of our team at once.
One of the best parts was sharing the space with our strategic partner, Ekumen. The team at Ekumen are experts in software development for the robotics industry. Through our partnership, InOrbit and Ekumen can offer the scale of a RobOps (DevOps for Robotics) platform and the flexibility of a dedicated team of developers who can build on top of the platform.
Our friends at Rover Robotics were also at the show. At the InOrbit booth, we demonstrated teleoperations of one of Rover’s modular robots through InOrbit Mission Control.
In addition, Florian moderated a panel on Robotics Infrastructure at Global Scale with some amazing panelists. There were some great insights on the challenges when robots first venture out of the lab into the field, and then the difficulties they found in scaling infrastructure for robot fleet management.
We were so busy we weren’t able to attend many of the great talks. One of our favorites was Prof. Ken Goldberg’s keynote, which not covered advances in robot grasping but also shared insights on cloud robotics, an opportunity that we actively pursuing at InOrbit.
The market for autonomous service robots is on the verge of a massive expansion. This has been largely driven by advancements in core technologies such as computer vision as well as lower cost of key hardware components and a standardized software stack.
Like we’ve seen with other technology waves such as cloud and mobile, venture capital is pouring into the robotics space, tripling in size from 2016 to 2017 and reaching +5B in 2017 by some estimates, which in turn is attracting more startups tackling an incrediblevarietyofproblems, from life-saving to mundane.
However there are still some big potholes on the road to widespread robot adoption. We cover 5 of them here; number 4 will not surprise you (if you are already scaling your robots). To learn more, join us at RoboBusiness on September 27, 2018 in Santa Clara, CA for our panel on Robotics infrastructure at Global Scale.
Robotics technology is fascinating, with many interesting technical challenges. However, the robotics companies that will succeed are those that solve a real business need. This requires specialization and single-minded focus on solving a specific use case, be it moving packages efficiently in a warehouse or hospital, scanning retail shelves or securing an office.
Spending time working on common infrastructure that every robotics company has to solve, such as secure, reliable communication with your robots, is not a great use of your limited resources, including your team’s attention.
Over the last few years, there has been great progress developing tools that help roboticists become more productive. This open source effort yielded platforms like ROS and tools like Gazebo, which have helped dramatically accelerate early-stage robot development.
However, once robots venture out into the real world, these tools have distinct limitations. For instance, recording ROS bags on production robots to collect diagnostics information presents a number of problems, from the risk of filling the robot’s hard-drive to sending large amounts of data over a flaky WiFi or costly LTE connection in order to pluck just a few useful insights. Even checking simple robot vitals such as battery load and HDD utilization become critical.
You may be able to manage a handful of robots by connecting to them one by one from your terminal. Once your fleet grows beyond that, this quickly unravels. Although roboticists rely heavily on these point-to-point connections with full access to the robot during development and initial testing, as the number of robots increases it can become a liability.
In fact, moving beyond SSH could almost be seen as a litmus test for the maturity of a robotics company that is getting serious about operations at scale. This is usually a painful transition; to quote a VP of Operations at a robotics company, there can be “much whining from engineers who are used to connecting directly” to a robot. The trade-off is increased accountability, security and robot management effectiveness.
Let’s face it: modern autonomous robots are essentially a mobile supercomputer with advanced data processing, running computer vision algorithms and other real-time calculations. A fleet of robots consists of distributed compute, storage and networking capabilities, just like a data center. However, in a data center the conditions in which machines operate are tightly controlled, with environmental, physical arrangement, power and connectivity having a high degree of reliability and consistency.
Robots, on the other hand, operate in uncontrolled environments, sometimes in far-flung locations, are usually mobile, have tight real-time computation constraints (eg to avoid obstacles), connect to unreliable networks and run the risk of running out of power in the middle of a task without a way to return to the charging station. And that’s without even getting into possible mechanical failures.
If you tried to design the least efficient data center in the world, you’d have to try hard to do worse than a fleet of autonomous robots. This means that robot management, including real-time monitoring, incident management and teleoperations, are much harder.
The service robot industry is still evolving and has yet to develop consistent best practices like software-as-a-service developers have created over the years. Even basic maintenance like updating software is still largely a high-touch, artisanal process. In order to reach the level of reliability and predictability required for massive adoption, a set of best practices will need to emerge. These will range from data protection and privacy to auditable operations.
At InOrbit, we’re working on addressing these problems. Our cloud-based robot management system is designed to give robotics developers the boost they need at lift-off and the ongoing operational support they need to scale.
We have developed unique capabilities such as adaptive diagnostics, real-time dashboard and analytics, and remote micro-interventions all built on secure, scalable infrastructure. InOrbit enables robotics companies and operators at all stages of deployment to focus on getting their robots to work and solving specific use cases.
Here are some examples for how we address these problems.
Our robot-side agent takes less than a minute to install and supports secure, reliable, bi-directional communication as well as common robot vitals out of the box. It can be easily extended to adjust to each robot’s unique needs.
With adaptive diagnostics, the agent responds to conditions on the robot to capture and upload the most relevant diagnostics information and makes it available immediately for remote analysis. Incident management with built-in integrations for PagerDuty and Slack allows teams to resolve issues as they emerge.
Secure operations are handled through a number of best practices, including role-based access control, encryption at rest and on the wire, renewable cryptographic key management, auditable policy-driven controls and context-aware robot resource management. We provide tools for non-technical operators from micro-interventions (small nudges to get the robot back to working autonomously) all the way to full teleops.
InOrbit Agents efficiently manage storage, compute and network resources to avoid interfering with the normal operation of the robot, balancing robot and cloud data aggregation and analysis. Hierarchical configuration management makes it possible to have consistent software deployments while also managing the unique settings for each robot deployment.
Our team has experience scaling SaaS products and working with dozens of robotics developers, from Willow Garage to the latest crop of RaaS solutions. As we continue to work with many of the leading companies in autonomous robotics, we are helping develop best practices for RobOps similar to the emergence of DevOps over the last 10 years.
We’d love to hear from you. Are you running into these issues? Are you ready to scale but don’t know how to get started? Have you built it yourself but found your team getting pulled in other directions so you can never get your tools to be as reliable as you need them to be?
Come see us at the RoboBusiness expo and we’ll help you get your robots InOrbit.
InOrbit, the platform for RobOps (DevOps for Robotics) is now generally available. After months of working closely with customers who joined our Early Access Program, we are ready to work with robotics companies that are growing their fleet of autonomous robots.
InOrbit is a secure SaaS platform that enables robotics companies to develop, deploy and operate autonomous robots at global scale. Our customers and partners work across all industries, from agriculture and hospitality to logistics and retail.
“We have been trying InOrbit for several months and can clearly see the business value it has for us in helping scale our robot operations. The combination of robust telemetry infrastructure and intuitive UX allows us to efficiently manage our growing robot fleet.”
Thavidu Ranatunga, CTO, Fellow Robots
“InOrbit helps customers developing on the Open Rover platform to set up fully autonomous robot applications with built-in monitoring and control, enabling non-technical users to manage their robots remotely.”
Adam Gettings, CEO, Rover Robotics
Out of the box, InOrbit provides a rich set of functionality, including:
Our platform is designed with extensibility in mind, allowing customers to easily integrate robot- and application-specific software. Additional functionality coming soon to the platform includes configuration management and predictive analytics.
We will be presenting our solution jointly with our partners at RoboBusiness on September 26-27, 2018. Come see us at the RoboBusiness expo and we’ll help you get your robots InOrbit.