DOD researches robot movements to improve dexterity

The Defense Department’s research and development wing is looking for ways to make ground robots more mobile and more dexterous when they manipulate objects. Robots have become invaluable in helping U.S. and allied forces find and disable explosives, but they are still very limited in travelling across rough terrain and in the fine reflexes needed to delicately handle objects.

The Defense Advanced Research Projects Agency’s Maximum Mobility and Manipulation (M3) program wants to greatly improve how robots do these tasks. It also seeks faster and more efficient ways to design and build prototype machines.

For mobility and manipulation, DARPA is asking companies to look at nature for examples. Besides land locomotion, the broad agency announcement also noted that other types of movement across many environments, such as water and air, would also be considered.

Different types of control mechanisms will also be examined. The program will initially look at methods using augmented teleoperation, or supervised autonomy. However, as the program progresses, more advanced types of autonomous techniques such as full autonomy, cognition, reasoning, high-level decision-making and decision-making under uncertain circumstances could be added. Basic low-level intelligence, the type needed to choose the best travel routes and avoid obstacles, are also key parts of the M3 program.

The M3 effort is built around four parallel tracks. The first looks at developing design tools that can quickly model, simulate and test a variety of robot designs. Track 2 seeks to develop new ways to manufacture robots through rapid prototyping techniques and novel assembly technologies.

Control for mobility methods are the focus of the third track. The program is looking for new ways for machines to vary and select their gate or movement rate based on the environment they’re in and to dynamically control their stability and body mass. DARPA expects researchers to look at how animals accomplish these tasks and apply similar methods to robotic movement.

Another important part of track three is variable compliance manipulation. Researchers and organizations will explore new, more flexible methods for manipulating objects. DARPA is also looking for manipulation methods to include haptic, or touch-based feedback that will allow a machine to automatically apply the right amount of strength or pressure to pick up and move an object without damaging it.

The final track consists of developing a technology demonstration prototype.