Acceptance, Trust Key for Future of Ground Robot Autonomy
The future of ground robot autonomy will hinge largely on trust and acceptance, according to a panel that spoke on the third and final day of AUVSI’s Unmanned Systems Defense.
“I see acceptance as being one of the big drivers going forward,” said Michael Bruch, Code 30 autonomy lead at the Office of Naval Research. “I think both on the civilian side, and on the military side, we’re seeing a lot more acceptance and interest in unmanned systems, and I think as much as technology, that’s going to drive the military going forward.”
Michael Wagner, CEO of Edge Case Research and a senior commercialization specialist at Carnegie Mellon University, said the only way acceptance will become reality is if people know the systems are reliable, which highlights the importance of testing.
“Part of that [acceptance] is knowing that these systems are safe and secure, and that’s been the focus of the research that my group has been doing,” said Wagner.
The panelists agreed there are a lot of opportunities on the horizon for autonomous systems, but there are obstacles that need to be hurdled to take the next steps forward.
Dr. Bob Sadowski, the Army’s chief roboticist, said he doesn’t see any large obstacles at hand, although he drew laughter when he warned that “vision without funding is a hallucination.”
Sadowski mentioned other obstacles, one of which is the need to turn unmanned systems from tools to teammates. But Sadowski made sure to emphasize that there is no such thing as a perfect solution, saying that “if we wait for the perfect solution, she’s [fellow panelist Camille Robbins] never going to be able to test it.”
Robbins, chief of the Autonomous Engineering Support Branch for the Unmanned Vehicle Division at the U.S. Army Aberdeen Test Center, said one of her main concerns is being involved from the beginning when it comes to the acquisition process, so that she and her team can get a full picture of the systems that they are dealing with.
“We have been working and are trying to focus more on integrating ourselves with S&T throughout the entire continuum of the acquisition life cycle,” she said. “The earlier we get involved, the more we understand that autonomy and how it makes decisions, and how we can interface with it.”
She also said the test community needs to accept “more approaches to testing, such as modeling and simulation.”
Panelist Carl Henshaw, a roboticist at Naval Research Laboratory, said deep learning algorithms pose a challenge in that “when you get into multi-modal deep learning … we are going to have systems that we know perform better than humans, but we’re not really going to understand what they are doing internally. So will have to be able to trust the results of the tests rather than the design.”
Sadowski echoed earlier speakers at Unmanned Systems Defense in saying that in the future, robots should take the first casualties in war, being first into “the breach,” and running point after that as well.
“Attrit the platform and not the crew,” and save their experience for later battles, Sadowski said.
Army Automated Trucks
In an effort to take soldiers out of harm’s way, the Army is looking to integrate automated trucks into their use, but there are several obstacles that stand in the way of that, according to Dr. Shawn C. McKay of RAND Corp.
A lot of the challenges that the Army is running into when it comes to the development of ATs are directly linked to the fact that a lot of environments where these vehicles would be used are anything but smooth and secure.
McKay said he believes the Army can take a lead from commercial, driverless platforms and use some of their strategies in an effort to create the safest vehicles, such as communications technology, human-machine interface improvements and sustainment and maintenance. But each of those areas has their own set of concerns.
In terms of communications, there is always the possibility of cyber threats. Commercial vehicles have faced those same issues, so there’s no reason to believe that vehicles used in warfare wouldn’t have the same threats, or worse.
With human-machine interface, there would be a need for a leader or remote guidance for automated trucks, and the tactical design requirements might constrain the usable space in cabs.
Finally, the mission availability of systems is a primary concern for sustainability and maintenance, which likely would require new competencies for Army maintainers and technicians.



