Skip to content

Box-DDP

Wolfgang Merkt requested to merge (removed):topic/wxm-boxddp into devel

Work-in-progress for Box-DDP. It works for some examples (quadruped walking), but not for others (jumping etc.).

Quadruped walking within control limits:

20191003_boxddp_walking_within_control_limits

Current issues:

  1. The Box-QP is a quick implementation and does allocate memory during runtime.
  2. The stopping criteria don't work right now. Walking is within limits within ~400 iterations, but it will happily continue optimising for many, many more.

Cost convergence plot for quadruped walking (log scale):

20191003_boxddp_walking_cost_log_scale

Edited by Wolfgang Merkt

Merge request reports