Most of my current work focuses on search-based motion planning for (aerial) vehicles in unstructured, cluttered environments. To find a collision-free and dynamically feasible trajectory for the robot, we search a graph of dynamically feasible motion primitives. In order to minimize computational complexity, I try to create optimally sparse planning graphs. These graphs allow us to trade off computation with planner completeness and optimality with a single parameter, the dispersion, which quantifies the largest regions of state space that are inaccessible from the planning graph. Most recently, we demonstrated our approach to achieve autonomous flight in a forest!

L. Jarin-Lipschitz, J. Paulos, R. Bjorkman, V. Kumar, "Dispersion-Minimizing Motion Primitives for Search-Based Motion Planning", ICRA 2021

L. Jarin-Lipschitz, X. Liu, Y. Tao, V.Kumar, "Experiments in Adaptive Replanning for Fast Autonomous Flight in Forests", accepted at ICRA 2022 (preprint)

Publication:

• I. D. Miller et al., "Mine Tunnel Exploration Using Multiple Quadrupedal Robots," in IEEE Robotics and Automation Letters, April 2020, (also presented at ICRA 2020)

Press:

• https://penntoday.upenn.edu/news/robots-rescue-underground-DARPA (that's me in the video in the white raincoat! Advice to past self: do not wear white clothing to a coal mine...)

About the challenge:

• https://www.subtchallenge.com/