At the recent Build conference in Seattle, Microsoft announced, in limited preview, an end-to-end toolchain to help developers and organizations build autonomous systems for their industries. The platform includes machine teaching tools and simulation technologies that enable intelligent robotic systems to complete tasks like running autonomous forklifts and robotic inspection platforms.
The platform leverages some existing Microsoft technologies, including Azure Internet of Things, Robot Operating System (ROS) for Windows, machine teaching technology that was acquired through the acquisition of Bonsai in June 2018 and Microsoft’s industry simulator technology called AirSim.
Machine teaching, not to be confused with machine learning, uses a complementary approach to solving AI problems by gaining knowledge from people rather than extracting knowledge solely from data. Jennifer Langston, an AI writer at Microsoft, explains:
Machine teaching relies on people’s expertise to break a problem into easier tasks and give machine learning models important clues about how to find a solution faster. It’s like teaching a child to hit a home run by first putting the ball on the tee, then tossing an underhand pitch and eventually moving on to fastballs.
Industrial automation is hardly a new phenomenon, however those systems differ from the ones that run autonomously. Traditional automation relies upon structured and repetitive processes as opposed to scenarios that involve variation or undefined end states. For example, consider an earthquake where a snakelike robot is able to crawl through debris or tight air pockets looking for people who may be trapped or dangerous environmental conditions.
Microsoft has partnered with Sarcos to bring intelligent capabilities to the Guardian S that allows the robot to move more autonomously by avoiding obstacles and providing the robot operator with more cycles to focus on pressing decisions. Kristi Martindale, executive vice president and chief marketing officer for Sarcos, explains:
In a real-world scenario, the operator would still play a role in guiding the robot. But if the Guardian S can sense its surroundings and perform all the intermediate motions to traverse stairs on its own, the operator can focus on assessing the scene and making more critical judgement calls.
Image source: https://www.sarcos.com/products/guardian-s/
Another important aspect to Microsoft’s autonomous systems platform is running simulations. It may not be practical, or feasible, for robots to learn on the job, making several expensive mistakes along the way. AirSim is an open source simulation platform, originally developed by Microsoft, which uses AI to teach drones, self-driving cars or robots in simulated environments.
Using simulations provides the ability to take a larger problem and break it down into many smaller tasks. Simulations also allow you to run many concurrent experiments which may reduce time to market. Ashish Kapoor, principal research manager at Microsoft, explains:
If I have the ability to spawn thousands of simulations at once and in each one the pedestrian crossing the street is different and the curve of the road is different, suddenly the AI system is able to gather much more diverse experience in a short amount of time. Azure gives us the ability to run these simulations at scale, which is really important.
One of the customers that Microsoft is working with on simulations is Toyota Material Handling, who is developing autonomous forklifts that “break the task down into sub-concepts that are simpler to learn and debug: navigating to the load, aligning with the pallet, picking it up, detecting other people and forklifts, delivering the pallet, returning to the charging station.”
Image source: (screenshot) https://www.youtube.com/watch?v=0yTDnJAqKow
For additional information on Microsoft's Autonomous Systems AI vision, please visit their website.
