Jumping over obstacles with MIT Cheetah 2

This paper presents a planning framework for jumping over obstacles with quadruped robots. The accomplishes via structured predictive control strategy that combines the use of heterogeneous simplified models different prediction time scales. A receding multi-horizon controller coordinates approach before jump using kinematic point-mass model. Consideration optimal value function horizons enables system to select an appropriate number steps take jumping. motion is then tailored sensed obstacle by solving nonlinear trajectory optimization problem. solution this problem online enabled exploiting analyticity flow map planar bounding template model under polynomial inputs. By combination models, MIT Cheetah 2 shown autonomously up 40 cm in height during high-speed bounding. Untethered results showcase ability method automatically adapt heights and placements single trial.