Embedded Files
Physical environments can provide opportunities for robots to exploit towards their locomotion goals. However, it is unclear how to even extract information about – much less exploit – these opportunities from physical properties (e.g., shape, size, distribution) of the environment, especially for non-flat, non-rigid, complex terrains. Our work seeks to develop models and framework that allow robots to intelligently elicit and select responses from their physical environments towards successful locomotion and navigation.
Research Projects
Research Projects
Our interaction models and framework allow bio-inspired robots to use their legs and tail to generate desired responses from cluttered obstacles and soft sand for effective locomotion and navigation.
Terrain responses characterization and manipulation
Terrain responses characterization and manipulation
Robots usually do not cope well with loose sand or boulder fields, primarily due to limited understanding of complex interactions between robot appendages and non-rigid, non-flat terrains. By designing automated terrain creation systems and performing systematic laboratory experiments, we develop interaction models and framework that allow robots to predict terrain responses and reason about interaction opportunities within their environments.
Selected Publications:
The dynamics of legged locomotion in heterogeneous terrain: universality in scattering and sensitivity to initial conditions, F. Qian and D. I. Goldman, Robotics: Science and Systems (RSS), Rome, Italy, 2015.
Walking and running on yielding and fluidizing ground, F. Qian, T. Zhang, C. Li, P. Masarati, A. Hoover, P. Birkmeyer, A. Pullin, R. S. Fearing, and D. I. Goldman, In Proceedings of Robotics: Science and Systems (RSS), Sydney, Australia, July 2012.
Robots using their legs as embodied sensors to help geoscientists map soil erodibility and fight desertification
Embodied sensing of environment responses for human-robot collaborative exploration
Embodied sensing of environment responses for human-robot collaborative exploration
Natural environments present complex spatial gradients and rich responses. For legged robots walking on sand dunes or through muddy environments, ground reaction forces felt from the legs can be as informative, or more informative, than visual or other sensing inputs for inferring environment properties and trends.
We develop direct-drive robots with embodied sensing capabilities, to help geoscientists collect environment data at every step. We also work closely with cognitive scientists to explore how human-robot teams can collaboratively adapt sampling strategies based on collected information.
Selected Publications:
Rapid in-situ characterization of soil erodibility with a field deployable robot, F. Qian, D. B. Lee, G. Nikolich, D. E. Koditschek, and D. J. Jerolmack, Journal of Geophysical Research: Earth Surface 124.5 (2019): 1261-1280.
Ground robotic measurement of aeolian processes, F. Qian, D. J. Jerolmack, N. Lancaster, G. Nikolich, P. Reverdy, S. F. Roberts, T. Shipley, R. S. Van Pelt, T. M. Zobeck, and D. E. Koditschek, Aeolian Research, 27 (2017): 1-11.
Different wave forms of body shape allow a snake-like robots to generate different paths from the same environment. Similarly, different leg coordination patterns allow a quadrupedal robot to robustly move towards different directions under repeated obstacle collisions.
Shape morphing and gait adaptation for obstacle-aided path selection
Shape morphing and gait adaptation for obstacle-aided path selection
Traditional navigation and planning methods largely rely on avoiding obstacles and large perturbations. We discovered that using different body shapes or leg coordination patterns (gaits), robots can actively select interaction opportunities from their environments to generate desired motion. By representing robot legs and body segments as "interaction opportunity selectors", we are creating legged and snake-like robots that can take advantage of obstacle collisions to effectively move through cluttered and perturbation-rich environments.
Selected Publications:
An obstacle disturbance selection framework: emergent robot steady states under repeated collisions, F. Qian and D. E. Koditschek, The Internationfal Journal of Robotics Research (2019): 0278364920935514.
Planning of Obstacle-aided Navigation for Multi-legged Robots using a Sampling-based Method over Directed Graphs, K. Chakraborty, H. Hu, M. D. Kvalheim, and F. Qian, IEEE Robotics and Automation Letters 7, no. 4 (2022): 8861-8868.
Page updated
Report abuse