Using Social Navigation Behaviors to Optimize AMR Performance

A summary of the Formant article”5 Takeaways from the Robotics Summit & Expo 2022″ featuring Zebra Robotics Automation.

Making it possible for robots and people to work together is a passion for Melonee Wise, VP-Robotics Automation at Zebra. As the founder of Fetch Robotics, the advanced autonomous mobile robot (AMR) widely deployed in warehouses, manufacturing operations and healthcare facilities, Melonee is also passionate about AMR safety.

In her presentation, “Why the Cloud Is a Force Multiplier for Robotics“ at the 2022 Robotics Summit and Expo, Wise shared the work she and her team are doing at Zebra Robotics Automation around the use of data and machine learning to optimize performance of AMRs, particularly in the field of robotic navigation. Working on social navigation behaviors, she and her team are focused on making it possible for robots to behave more like people and make people more comfortable around those robots. For example, when a person is walking down the right side of an aisleway or a thoroughfare inside of a warehouse, these advances ensure that the robot will do the same thing so that they’re out of the way of traffic. When a person comes head-on, the robot will go to the right and the person will go to the right as well.

Some of the highlights of Melonee’s work

  • Like people, robots don’t always behave as one might expect them to.
  • In developing social navigation behaviors for robots, geography matters – e.g., robots in the UK need to model the left-sided world, as opposed to a right-sided one as in the U.S.
  • The next frontier in robotic innovation will be made possible with the ability to use data to build and extend social navigation behaviors and provide valuable insights to customers.

After modeling hundreds of scenarios, we found robots don’t always behave how you expect. Using these findings, we rounded off the sharp edges of robot-human interactions, such as ensuring that a robot defaults to moving to the right when moving directly at a human. Through pulling tons of simple lidar data off devices, we introduced behavioral changes that produced significant improvements in fleet performance.

Melonee Wise, VP Robotics Automation, Zebra Technologies

Read the full article on Formant here.

More Articles from the Fetch Blog