Safe Trajectory Generation of Robotic End-Effector Poses for RPOD Emulation Using Bézier Curves

Jason K. Schmidt, Space Dynamics Laboratory
Francisco J. Franquiz, Space Dynamics Laboratory
David C. Huish, Space Dynamics Laboratory

Abstract

System-level operational testing of space vehicles and payloads designed to execute rendezvous, proximity operations, and docking (RPOD) missions are currently tested using hardware-in-the-loop and processor-in-the-loop simulations. Despite the advantages offered by such test platforms, they rely on models, propagators, and scene generators to approximate synthetic on-orbit scenarios. This work details the hardware, software, and algorithmic implementation of a testbed designed to replicate on-orbit dynamics and illumination, as seen by sensors and payloads onboard spacecraft. The testbed, owned by the Space Dynamics Laboratory (SDL) and called the Hardware-in-the-Loop Operations Laboratory (HALO Lab), enables technology demonstrations for RPOD hardware and software. It also provides a realistic environment for the training of mission controllers.

The testbed hardware consists of two industrial robot manipulators, each with six degrees of freedom, placed on linear tracks arranged in a T configuration within a 17 m × 35 m × 5.5 m volume. The orientation of the tracks allows for emulation of most relative pose trajectories in an RPOD scenario by distributing the motion between the robots. The robots are commanded in real time by a custom server/software controller developed by SDL.

The robots are capable of very high accelerations that could damage sensitive instruments. The controller must command the location and attitude of the robot end-effector at 250 Hz over trajectories that maintain the safety of mounted hardware as well as personnel and equipment in the room. The commands must be realizable and follow arbitrary customer-supplied trajectories as closely as possible. Cartesian and quaternion Bézier curves are used to interpolate customer-provided trajectories. Bézier curves are well suited for smoothly interpolating sparse Cartesian position data. Relatively recent developments have extended the concept of Bézier curves to quaternions, allowing the same techniques to be applied to sparse attitude data. A primer on arbitrary order Cartesian and quaternion Bézier curves is presented.

Customer trajectories are transformed into the HALO Lab coordinate frame, and Bézier curves are applied to the problem of smoothly interpolating arbitrary position and attitude waypoints in the facility. Next, an algorithm is presented that iteratively modifies the provided trajectory (which may not be safe) to satisfy position, velocity, and acceleration constraints as well as attitude, angular rate, and angular acceleration constraints while still following the requested trajectory as closely as possible.

Finally, real-world results are presented including the quality of the trajectories generated by the algorithms and the control biases of the robot following the trajectory.

 
Aug 9th, 5:29 PM

Safe Trajectory Generation of Robotic End-Effector Poses for RPOD Emulation Using Bézier Curves

Utah State University, Logan, UT

System-level operational testing of space vehicles and payloads designed to execute rendezvous, proximity operations, and docking (RPOD) missions are currently tested using hardware-in-the-loop and processor-in-the-loop simulations. Despite the advantages offered by such test platforms, they rely on models, propagators, and scene generators to approximate synthetic on-orbit scenarios. This work details the hardware, software, and algorithmic implementation of a testbed designed to replicate on-orbit dynamics and illumination, as seen by sensors and payloads onboard spacecraft. The testbed, owned by the Space Dynamics Laboratory (SDL) and called the Hardware-in-the-Loop Operations Laboratory (HALO Lab), enables technology demonstrations for RPOD hardware and software. It also provides a realistic environment for the training of mission controllers.

The testbed hardware consists of two industrial robot manipulators, each with six degrees of freedom, placed on linear tracks arranged in a T configuration within a 17 m × 35 m × 5.5 m volume. The orientation of the tracks allows for emulation of most relative pose trajectories in an RPOD scenario by distributing the motion between the robots. The robots are commanded in real time by a custom server/software controller developed by SDL.

The robots are capable of very high accelerations that could damage sensitive instruments. The controller must command the location and attitude of the robot end-effector at 250 Hz over trajectories that maintain the safety of mounted hardware as well as personnel and equipment in the room. The commands must be realizable and follow arbitrary customer-supplied trajectories as closely as possible. Cartesian and quaternion Bézier curves are used to interpolate customer-provided trajectories. Bézier curves are well suited for smoothly interpolating sparse Cartesian position data. Relatively recent developments have extended the concept of Bézier curves to quaternions, allowing the same techniques to be applied to sparse attitude data. A primer on arbitrary order Cartesian and quaternion Bézier curves is presented.

Customer trajectories are transformed into the HALO Lab coordinate frame, and Bézier curves are applied to the problem of smoothly interpolating arbitrary position and attitude waypoints in the facility. Next, an algorithm is presented that iteratively modifies the provided trajectory (which may not be safe) to satisfy position, velocity, and acceleration constraints as well as attitude, angular rate, and angular acceleration constraints while still following the requested trajectory as closely as possible.

Finally, real-world results are presented including the quality of the trajectories generated by the algorithms and the control biases of the robot following the trajectory.