KinematicTrajectoryOptimization seems to fail arbitrarily on some inputs when using a DurationCost and AccelerationBounds, even when setting the AccelerationBounds far outside the actual acceleration of the trajectory (or even setting them to infinity).

The goal is to prevent the end effector from reaching a certain area

| drake 1.27.0 | python 3.8 | OS: ubuntu 22.04 | pycharm | ROS 2 Humble | I’m trying to simulate use of a UR10e with the robotiq 85 gripper for simple pick and place tasks. I’m using these models – urdfs that are combined using sdf – then parsing them with this code snippet: […]