Virtual model control for quadrupedal trunk stabilization

Legged robots have the advantage of being able to access a large variety of terrain types. In unstructured environments, where only a number of discrete footholds are possible (disaster sites, construction sites, forests, etc.), legged quadrupedal systems can utilize static stability crawl gaits or dynamic stability –often ZMP-based– gaits. In environments where smooth, continuous support is available (flats, fields, roads, etc.), where exact foot placement is not crucial for the success of the behaviour, legged systems can utilize a variety of more dynamic gaits, e.g. trotting, galloping. In this work we focus on a trotting controller for the hydraulically actuated quadruped HyQ [1]. The trot is a symmetrical gait in which diagonally opposite legs swing in unison. This allows a desirable division of the total force that the leg actuators should be able to provide when supporting the weight of the robot, when thrusting and when receiving impact forces at touchdown. In addition, the center of gravity of the system is on average kept above or very close to the line of support that the stance legs define. This leads to a more stable gait with respect to the robot’s attitude, in contrast to bounding and pacing. In practice implementing trotting controllers in real world quadrupeds has proven a challenging task, especially for large robots such as the HyQ (∼75kg). This is due to a number of factors, the most significant two being the need for compliance when legs are touching down, both for impact absorption and surface uncertainty handling, and the need for accurate control, both for foot placement during swing and attitude stabilization during the stance phase of each leg pair. In this extended abstract we present our efforts towards a trotting controller with an explicit trunk stabilization goal. Trunk stabilization serves to reduce oscillations of the body in the vertical direction, that often lead to increased load for the leg actuators, while also allows for more accurate foot placement of the swing legs. In addition, trunk stability is crucial for the successful operation of higher order onboard sensory modalities, such as lidars or stereo cameras, that are essential for perceptual processes as mapping and localization.