Robotics · 2026
Franka robot motion planning
Motion planning and optimization for a seven-degree-of-freedom Franka Emika robot.
Problem
As part of a multi-team robotics project, my team developed the motion-planning for a Franka Emika Panda robotic arm responsible for dispensing candy into a cup. The program needed to take over control from another team after they received the cup, position it beneath the dispenser, actuate the dispenser lever, retrieve the filled cup, and return it back to the neutral position for its next task.
Engineering Approach
We decomposed the task into a sequence of 14 waypoints representing the major stages of the operation. Each waypoint represented a homogeneous transformation relative to the robot’s base frame. The end-effector orientation was constrained to keep the cup upright throughout the motion.
We then used inverse kinematics to convert the desired end-effector poses into joint configurations, implementing joint-space planning.
To make the system adaptable to different setups, the system was designed to reference AprilTags to determine waypoint locations. The motion-planning was initially tested in Gazebo simulation before being validated on the physical Panda arm. During hardware testing, we adjusted the waypoint geometry based on the robot’s physical motion until the Panda could complete a reliable dispensing sequence.
The project required coordinating with multiple teams responsible for cup handling, scheduling, and world tracking. We established a shared neutral position and gripper width to ensure reliable handoffs between robot tasks.
Engineering Takeaways
The project provided hands-on experience in robotic motion planning and sim-to-real validation. It demonstrated the importance of designing motion-planning algorithms around the capabilities and limitations of physical hardware. The system prioritized simple and repeatable motion over more complex motions, to ensure the Panda could reliably complete its task within the project’s time and integration constraints.
Tools/Methods
Python, Gazebo, Franka Emika Panda, inverse/forward kinematics, joint-space planning, trajectory planning