Balancing
I implemented a robust balance control system for the quadruped robot utilizing a 6-DOF IMU that measured linear accelerations and angular velocities across all axes. The system employed complementary filters to fuse accelerometer and gyroscope data for accurate state estimation, while a PID controller continuously adjusted leg positions to maintain stability. Real-time center of mass calculations combined with IMU feedback enabled dynamic balance adjustments and gravity compensation.
Mechanical Design
For the quadruped robot's mechanical design, I prioritized cost-efficiency and rapid prototyping while maintaining functionality. The structure utilized affordable servos augmented with elastic elements to improve torque characteristics while keeping the parts cost low. I optimized the 3D-printed components for minimal material usage and print time, incorporating strategic infill patterns and support placement. The frame design balanced structural integrity with production speed, featuring modular components that could be easily reprinted and replaced. This approach resulted in a capable quadruped platform that achieved stable locomotion while keeping manufacturing costs and production time low.
Locomotion
The locomotion system combined inverse kinematics with an open-loop trotting trajectory generator to achieve dynamic walking. Foot positions followed pre-computed polynomial curves, while the IMU feedback enabled real-time foot adaptation to maintain stability on uneven terrain. The controller adjusted step height and touchdown positions based on detected roll and pitch angles, compensating for disturbances during the gait cycle. This approach achieved reliable trotting without complex algorithms, allowing all the locomotion programming to be ran on an Arduino.
Quadruped robot maintains balance on a dynamically tilting platform, demonstrating real-time IMU-based adaptation of foot positions and posture to counteract changing surface angles
3D render of the robot dog limbs, frame, and servos
Quadruped executes a trotting gait, with each foot following polynomial trajectories for ground clearance and touchdown, while inverse kinematics ensure correct leg positioning throughout the stride