Track manipulation, locomotion and embodied systems.
Search papers, labs, and topics across Lattice.
This paper couples a recurrent proprioceptive 3D force estimator with an RL locomotion policy, predicting explicit metric forces in newtons from joint history to simultaneously accommodate payloads and interpret leash-guidance vectors. In sustained-load simulation benchmarks, the approach lowers base-orientation errors by 46.5% and elevates survival rates from 68.3% to 94.6% over domain-randomized baselines with 1.44–2.83 N force RMSE. Real-world Unitree Go1 trials validate the shared interface across stationary sensing, leash-driven guidance, and locomotion under an out-of-distribution 8.5 kg payload.
Pure proprioceptive history can double as an accurate 3D metric force sensor, letting quadrupeds navigate leash guidance and carry payloads exceeding training limits without dedicated load cells.
Payload forces must be accommodated during locomotion, while leash forces can specify desired motion. We investigate whether a shared three-dimensional force estimate in newtons, inferred from proprioceptive history under sustained loading, can support both tasks. An estimator and locomotion policy are jointly trained with supervised force and velocity outputs and learned latent context. The estimated force conditions locomotion and additionally generates planar-velocity and yaw-rate commands for leash guidance through an analytical map. In sustained-force simulation sweeps, temporal means of componentwise force root mean square error range from 1.44 to 2.83\,N. Compared with a domain-randomized baseline, the framework reduces velocity-tracking and base-orientation error scores by 21.6\% and 46.5\%, respectively, and increases mean survival from 68.29\% to 94.60\% in separate sustained-force tests. Unitree Go1 experiments demonstrate stationary vertical and horizontal force estimation, locomotion with an 8.5\,kg payload whose weight exceeds the 70\,N training force limit, and leash guidance using the same force-estimation interface.