Walking Robot
Multi-legged robot with coordinated gait control and real-time sensor feedback implemented from scratch.
Overview
A multi-legged walking robot with a custom gait controller, inverse kinematics solver, and real-time sensor feedback loop. Built from scratch — hardware, electronics, and software.
Locomotion Algorithm
Implemented a central pattern generator (CPG) approach to produce stable gaits. The controller runs at 100 Hz and generates joint angle trajectories for each leg in real time.
Gait Modes
- Tripod gait: Three legs in contact at all times — fastest stable gait
- Wave gait: Conservative, one leg lifted at a time — maximum stability
- Adaptive mode: Switches based on terrain sensor feedback
Kinematics
Each leg is a 3-DOF serial linkage. The inverse kinematics solver maps desired foot positions to joint angles using a geometric closed-form solution, avoiding singularities at workspace boundaries.
Software Stack
- ROS (Robot Operating System) for inter-process communication
- Python for the gait controller and kinematics
- Arduino for low-level servo control
- Real-time sensor fusion using IMU data
Hardware
- Custom laser-cut acrylic frame
- 18 servo motors (3 per leg × 6 legs)
- Raspberry Pi for on-board compute
- MPU-6050 IMU for orientation estimation
Results
Successfully demonstrated stable walking on flat terrain and recovery from moderate perturbations. The adaptive gait switch reduced stumbling events by roughly 60% compared to fixed-gait operation.