This one is my bachelor's thesis project. Objectives of the whole project are listed below:
1- Designing a 4-bar mechanism for knee joint (Optimized with GA)
2- Simulating Robot (Choreonoid and Gazebo)
3- Design Control loop for dynamic balancing and robot speed control
4- State estimation for robot floating base velocity
To the left, you can see Gazebo simulation of the robot.
For the first part using Python programming and analytic solution of systems kinematic and kinetics (Newton method), a cost function including motor power and motor torque has been achieved. Then using the Genetic Algorithm optimum length for mechanism links has been evaluated. To the right, you can see GA progress.
After taking care of simulation using a CAD model drawn with SolidWorks, a ROS Package has been created for gazebo simulation. Another package has been designed for the main control loop. Both Packages were implemented in C++, and the same library with minor changes has been used for Choreonoid simulation. Below, you can see the block diagram of the robot control loop. The primary stabilizer in this system is the Linear Quadratic Regulator (LQR).
Robots kinematic has been kept as simple as possible. Inverse Kinematic analytic and geometric solutions have been used for both the 4-bar linkage system of the knee mechanism and also for the leg configuration.
Forward Movement at speed of 0.25 m/s
2D Motion (Turn+Forward)
Here a preliminary draft of a paper about this project is available. Note that this is a working paper that has not been submitted yet, so the final version may vary.
For future researches, I am considering the jumping and hopping motions of wheeled-legged robots. To the right, you can see the execution of simple jumping using pre-planned trajectory for robot length which has been derived using C2 polynomial trajectories.