This Broadening Participation Research Initiation Grant in Engineering (BRIGE) provides funding to enhance recent biological hypotheses regarding the control of high-speed legged locomotion over uneven terrain for application in robotics. A simulation model will be used to thoroughly test the hypotheses in ways not achievable in biological labs, such as lengthening limb segments or redistributing mass, to make critical enhancements. At high running speeds, legs utilize passive compliant elements such as tendons and ligaments to store and return energy during the stance phase of the leg cycle. To achieve robust locomotion, the effort of these passive elements must be combined with active feedback response from the muscles to overcome the pervasive uncertainty inherent to natural terrain. Two hypotheses derived from biological evidence, independently addressing the passive and active elements of locomotion, will be combined and tested for robustness, practicality and generality using full 3D models of biped and quadruped systems.
Legs appear in many forms, vary widely in function and can easily outperform man-made wheeled and tracked vehicles on uneven terrain, which makes the replication of leg functionality and the associated mobility very desirable for robotics systems. A robot that could achieve animal-like robust and agile movements, combined with the ability to work in remote and hazardous environments, would be valuable for time-critical search and rescue, planetary exploration, in vivo drug delivery, military reconnaissance, prostheses, hazardous waste cleanup, home service, and a host of other applications. Despite the great potential of legged systems, no biomorphic legged system is currently operational in a natural environment. This BRIGE grant provides funding to potentially enable improved robot locomotion using biological research as the foundation.
Legged systems will soon play a significant role during time critical search and rescue, planetary exploration and many daily tasks in the home. At this time, however, legged robots are limited in their ability to maneuver over uneven terrain. This has been the focus of this work, focussing separately on walking and running systems. For walking, an experimental 6-legged system with sprawled posture (like a cockroach) was developed to pursue the most effective use of force feedback. Humans and animals can rapidly and reliably determine if and how hard each leg is pressing on the ground and use this information to make decisions about stability and propulsion. This kind of information has been difficult to acquire for robotic systems until now. The Priniciple Investigator invented a novel mechanism for determining foot force continuously thoughout the walking step, even when the foot encounters terrain much higher or lower than expected. The PI's university has filed a provisional patent for this work and has licensed the technology to two companies, one interested in military usage and the other for advanced hobby and toy development. For running, the PI pursued the notion of periodic running as a means to simplify the control algorithm and reduce sensor dependence on uneven terrain. Imposing periodicity over uneven terrains suggests that regardless of the terrain height, the system will complete each stride in a fixed amount of time. It has been observed that animals running at high speed rely less on sensor feedback and more on clock-based circuitry, likely because of the difficulty in accurately sensing body state during the harsh dynamics occuring at high speeds. This approach was applied to a biped running system, and it was found that stiffening and retracting each leg immediately before it impacts the ground can eliminate the need to measure body state during the stride. It has long been known that horses execute this early retraction at high speeds, and this work has revealed a potential reason why. Together, the work on walking and running both serve to push the field of legged robots toward systems usable in complex environments, where wheeled and treaded systems have the most difficulty.