Login
iscover
The Machine Common Sense (MCS) program at DARPA has achieved a marked improvement in robotic performance, with computational models designed to enable the robots to continually learn from experience.
To continue reading the rest of this article, please log in.
Create free account to get unlimited news articles and more!
According to the Defense Advanced Research Projects Agency (DARPA), the MCS program has improved the capabilities and performance of robots through models that enable learning “just as infants must learn from experience”.
This has been achieved by using computational models that resemble a childlike cognition of objects (intuitive physics), agents (intentional actors), and places (spatial navigation).
It has been confirmed that over recent tests, that the robots have been able to demonstrate improvements in grasping objects, overcoming obstacles and changing speeds to meet the needs of a terrain.
“These experiments are important milestones that get us closer to building and fielding robust robotic systems with generalised movement capabilities,” Dr Howard Shrobe, MCS program manager in DARPA’s Information Innovation Office, said.
“The prototype systems don’t need large sensor suites to deal with unexpected situations likely to occur in the real world.”
In a release, DARPA illustrated that a quadruped robot was able to adapt to different terrain through a rapid motor adaption (RMA) algorithm that uses proprioceptive feedback.
The robot was able to thus navigate across both physical and virtual domains.
“The algorithm is trained completely in simulation without using any domain knowledge-like reference trajectories or predefined foot trajectory generators and is deployed without any fine-tuning,” the release confirmed.
Meanwhile, researchers from Oregon State illustrated how robots could carry loads with proprioceptive feedback, even learning how to demonstrate common sense behaviours to adapt to the environment.
MCS researchers are concurrently building scalable, machine-authored, symbolic knowledge bases that are designed to provide a better representation of the global environment.
“By focusing on commonsense, we are creating the possibility for systems to have the flexibility of human learning and the breadth of human knowledge,” Shrobe said.
“Fusing this knowledge with advanced robotics could result in highly capable, mission-critical systems that humans will want to have as partners.”
[Related: New DARPA program targets greater AI flexibility in battlespace]
