Crawling Faster, Clambering Higher

Crawling Faster Goldman Juntao Publication

Legged robots capable of traversing difficult terrain and uncertain environments could be revolutionary in applications from defense to mining to disaster search and rescue. Research into the development of motion controls for bipedal and quadrupedal robots has made great strides recently. Tests in highly controlled environments for bipedal robots and varied terrain in quadrupeds show promise, but the costs and complexity required to equip these robots with the sensors needed to navigate create a huge barrier to deployment at scale. In contrast, low-profile multilegged robots with redundant contacts eliminate the need for costly visual and LIDAR systems and are poised to be deployed commercially in the agricultural sector.

These multilegged locomoting systems, though less complex and costly, come with their own technological challenges that impact speed and vertical maneuverability due to the robotic design’s high degree degrees of freedom and visual sensing limitations due to height in relation to environment. To address these challenges, Juntao He, a PhD student in the group of Daniel Goldman, Professor in the School of Physics at Georgia Tech led a pair of research papers that paves the way to make these bots able to move faster and climb higher in challenging environments. Working with Baxi Chong, now assistant professor at The Pennsylvania State University as well as others in Goldman’s lab in a multidisciplinary collaboration to improve these cost-effective little bots.

To enhance speed on varied terrain, the researchers used a multilegged segmented robot equipped with 3 motors for pitch and yaw and leg tip mounted force sensors onboard each segment. Inspired by the movement of centipedes, the team added vertical body undulation coordinated with horizontal undulation and leg stepping. The additional vertical movement mitigates the environmental elements that impact forward motion, allowing the robot to move across multiple surfaces without a loss of speed.

The many-legged robot demonstrates impressive 2.5D mobility in unstructured environments with minimal sensing. What’s next? Our goal is to integrate greater intelligence into the robot, enabling it to make decisions and navigate effectively
in the open world. - Juntao He

To enable greater vertical obstacle navigation, Goldman’s team used the same robotic setup with the addition of tactile antenna to investigate impediments and a control system that integrates data from the antenna and the force sensors on the legs. This integrated data prompts the robot to adjust head placement and optimize the vertical undulation waves to climb the probed object. Using this efficient sensor system, the team’s robot reliably scaled obstacles five times its height in a controlled laboratory setting and performed equally well in outdoor testing. These scientific discoveries made by Juntao and other contributors are already being commercialized by Goldman’s startup Ground Control Robotics, inc.*

Publications Referenced
Probabilistic Approach to Feedback Control Enhances Multilegged Locomotion on Rugged Landscapes
Tactile sensing enables vertical obstacle negotiation for elongate many-legged robots

*Disclaimer: Daniel Goldman has an equity interest in Ground Control Robotics, Inc (GCR). GCR Inc. develops robots for locomotion in complex environments. GCR Inc. may potentially benefit financially from the research findings on locomoting systems presented here.

Institute for Neuroscience, Neurotechnology, and Society

Crawling Faster, Clambering Higher

Georgia Institute of Technology