Software companies, especially ones that can help other companies build products, often use the “Build or Buy” phrase to explain why companies should purchase software instead of building it themselves. This is no different within the robotics software space, where companies are offering to help companies save time and money in their robotics development. We’ve done this ourselves.
With the way robots are presented in popular culture, it’s easy to overlook how complex they are and how much goes into making a robot perform to meet our high expectations. We’re still in the early stages of how robots will revolutionize various industries, but the brilliant people solving problems to make that possible are making great strides.
On the software side, the lines are blurred between making the robot extremely reliable on its own and ensuring the performance of robot fleets in production settings. With all the unique challenges of deploying a new robot, it’s easy to lose sight of how much the operational workflow has in common with existing solutions that enable companies to scale their own fleets. Reinventing yet another wheel might not be the right choice.
RobOps addresses unique challenges beyond traditional DevOps
Savvy companies building out their robotics infrastructure recognize that existing cloud practices and services can handle many of the vital challenges in RobOps (Robot Operations). The time-tested approaches of DevOps provide a great foundation for the cycle of continuous improvement required for a highly reliable fleet. Even exceptionally designed robots are going to have autonomy exceptions that require real-time intervention and can inform the processes to avoid repeated failures. Robots that operate impressively in the lab are usually a different story in the field, often far from headquarters and in uncontrolled environments..
Autonomous mobile robots (AMRs) are especially prone to such exceptions. Considerations around each robot potentially generating terabytes of data from its sensors, moving through spaces with often compromised connectivity, and knowing how to respond to real and perceived obstacles are just a few of the challenges that arise for AMRs. But robots are normally quite robust once they are in production, the balancing act is to optimize the data management to support exactly what’s needed for informed intervention, when required, without incurring high costs for transmission and storage of unnecessary data.
When facing the “Build or Buy” question, the good news is that there are off-the-shelf integrated RobOps solutions available for all of the areas not typically found on a robot provider’s engineering team. The best of these offer extensible platforms that allow engineers and others at the company to customize and extend tools to meet specific needs and fill the gaps.
At that point it’s no longer Build or Buy, but Build AND Buy. Instead of being a company that recognizes that these key priorities need attention but doesn’t have capacity to address them, winners have a plan. They realize that trying to build too much of the infrastructure internally, plus the effort to maintain it over time, will slow down critical progress on core areas, and thus time to market.
For example, roboticists are often experts in improving the navigation stack and other subsystems. But alerting and empowering less technical and potentially remote operators to take control around exceptions is generally not a primary concern. Furthermore IT skills such as configuration management and security are critical requirements not typically seen on a robotics engineering team.
Similarly, while a mission management interface may be well covered, orchestration with other robots or enterprise software (such as warehouse management systems) in a heterogeneous environment is rarely on the radar of any but the most seasoned robotics teams. Front-end tools that monitor and visualize robots for internal team members or operators, tailored to the range of roles that need to track or take action on the fleet, are giant pieces of the puzzle. Streamlining the workflow between operators to avoid conflicts when a robot needs attention is often overlooked until it becomes a visible problem. Even measuring and compiling the analytics to support continuous improvement is often missed by initial robotics development teams.
Optimizing robots, but also people
Many companies building AMRs give them the ability to optimize workflows within a work location, yet those building the robots often don’t optimize their own workforce. Engineers, if left to their own devices, will try to solve problems many times over, often delaying projects.
There are clear advantages to the Buy side of the decision, such as not having to build capabilities from scratch and not having to maintain or expand them. In addition, deep considerations around security and scalability, or even fleet management, come into play, where a company may not have the required expertise.
Too often, companies feel compelled to build more than they really need to, underestimating the cost and complexity of developing, maintaining, and updating a homegrown solution for RobOps.. By being deliberate in partitioning the development with a well-designed RobOps platform that lets you build exactly the pieces that differentiate your solution, and focus on lightweight configuration and integration for the rest, you optimize your team and make your product shine.
The InOrbit platform has been crafted over several years to meet these common challenges from the ground up, leveraging deep experience deploying robotic systems and the best of DevOps practices. Flexible APIs and embeddable widgets offer both programmatic and packaged tools that companies can choose to accelerate time to market. To discover how InOrbit can help your team streamline development for scalable operations at any stage, schedule a demo.