Fall 2015 I took the Sangbae Kim's 2.740 Bio-inspired Robotics course in the department of mechanical engineering at MIT. The primary point of the course it to learn how to look for inspiration from natural systems. The secondary point of the class is to exploit the dynamics of a system by analyzing system kinematics. The key technical learning was how to simulate system dynamics, and how to use an impedance based controller, rather than a simple PID loop.

A few of the labs utlized a 3D printed, two motor, four-bar linkage robotic leg. The idea was great, but the mechanical design and implementation with 3d printed parts was full of friction, and stiction; it was difficult to tell if we were feeling the programmed spring and damping current controllers or just the friction of the system. So, I decided to design and build a clean piece of hardware with proper bearings.

For simulation of this experiment we worked out the kinematics of the system, and built a Matlab model.

For experimentation we wrote a motor controller on an ARM Cortex M4 microcontroller platform. For simplicity we utilized the MBED platform which is essentially a set of arduino libraries for 32-bit platforms.

Below are some links to the files used for these experiments.

• Experiment_Lab3_template.m
• keypoints_leg.m
• RunExperiment.m
• lab 3.pdf
• 3D CAD Files (Solidworks 2015)

The first step is to get the motor controller working. Rather than control position like traditional robotic platforms we want to control torque for our dynamic system. This means we needed to relate real-world torque output to input current. Some initial tests of the performance of a current and position controller operating are in the video below.