Simba GaoMECHANICAL / MECHATRONICS
← All projects

ROBOT MANIPULATION / SIMULATED BARRETT WAM

7-DOF WAM Manipulation and Control

I combined collision-aware planning and joint-torque control for a seven-axis arm, with torque commands communicated at 500 Hz.

My roleAutonomous manipulation algorithms, trajectory planning, and integration of joint-torque commands with state feedback.

  • Motion planning
  • Joint-torque control
  • State feedback
  • Robotic manipulation
Simulated Barrett WAM arm above a table with objects, a bin, and a raised obstacle.
Simulated manipulation workspace
7 DOFArm degrees of freedom
500 HzTorque-command communication rate

01 / ENGINEERING PROBLEM

Plan and execute motion around clutter

The simulated arm needed to move objects through a workspace containing obstacles. Planning a usable path was only part of the task: execution also needed joint-torque commands and continuing state feedback. I integrated these parts into a closed-loop manipulation workflow.

02 / METHODS & ANALYSIS

From a planned path to controlled motion

Frames and excerpt from the supplied replay
Click images to enlarge

01

Plan through the manipulation workspace

I developed collision-aware trajectories for the arm in the cluttered scene. These frames show different poses from the same replay; the geometry of the bin, barrier, and objects constrained the motion.

02

Connect the trajectory to torque control

  1. PlanCollision-aware trajectory
  2. CommandJoint-torque updates · 500 Hz
  3. SimulateSeven-axis arm dynamics
  4. Feed backRobot state returns to the control loop

I combined planned motion with joint-torque commands and robot-state feedback. Torque commands were communicated at 500 Hz, while state feedback connected execution to the planned motion.

03 / RESULTS & OUTCOME

Closed-loop manipulation replay

SIMULATED ROBOT · HARDWARE UNTESTED
26 s excerpt at original speed · cropped to the simulation viewport.

The supplied replay demonstrates the integrated planning-and-control workflow moving the arm around the tabletop scene. It supports a qualitative manipulation result; repeatability, collision-clearance margins, and physical-robot performance were not quantified in the supplied materials.