BSc Thesis:
Bio-Inspired Lightweight Soft Robotic Gripper
Designed, fabricated, and programmed the control architecture for a bio-inspired, antagonistically tendon-driven robotic end-effector.
Software Architecture & C/C++ Development
- The software was built on arduino using dynamixel motors and .
- A non-blocking state machine was used to handle real-time transitions, motor control, and continuous ADC sensor polling without execution delays.
- To be able to communicate with multiple motors simultaneously, which was very important for this project, I implemented the Dynamixel SyncWrite function. By using a custom buffer over Half-Duplex UART TTL and batching data, I achieved a 1.74 ms motor read/response cycle.
- Bridging Mechanics & Logic: Integrated and calibrated strain gauge load cells via an HX711 ADC (10 SPS), translating raw mV changes from a Wheatstone bridge into actionable digital force feedback to evaluate grip and lift success.
Soft Robotics & Control Challenges
- Antagonistic Tendon Control: To give the soft fingers stiffness, and to make the fingers curl in a spiral like an octopus, we implemented an antagonistic tendon control system. In the same way a human finger has opposing flexor and extensor muscles, the gripper has two sets of tendons, for each of its two fingers. Pulling these tendons antogonistiacally (at the same time) created additional strength and stiffness. Additionally we utilized the concept of the capstan equation (the equation for tendon friction), we made the fingers curl around the objects being gripped.
- Compliance & Collision: Shifted away from positional motor control, by implementing a feedforward PWM actuation strategy, to make the robot press only with a certain force, paired with current feedback to update the state machine. This allowed the software to "feel" initial collisions and adapt to the unpredictable deformations of soft robotic grasping.