Prototyping a Quadruped Robot Leg with Brushless Actuators

Added:

Quasi-Direct Drive (QDD) Actuation

While ball screws and high-reduction transmissions offer high force and precision, they suffer from low backdrivability, high reflected inertia, and poor shock tolerance. When a quadruped robot runs or jumps, its legs experience intense, high-frequency impact forces. In rigid transmission systems like ball screws, these impact forces are transmitted directly back into the drivetrain, which can cause catastrophic mechanical failure. To address this, modern quadruped design heavily favors Quasi-Direct Drive (QDD) actuation—using high-torque, low-gear-ratio planetary gears or direct-drive motors. QDD maximizes 'backdrivability,' allowing the leg to behave like a physical spring that naturally absorbs impacts. This compliance protects the robot's structure and enables proprioceptive force control (sensing ground contact directly through motor current) without requiring heavy, expensive force/torque sensors. Consequently, for dynamic and agile locomotion, QDD is widely considered superior to high-reduction linear transmission systems.

Fundamentals of Brushless DC (BLDC) motors and Field-Oriented Control (FOC) for precise torque and speed management.
Basic mechanical transmission principles, specifically how ball screws convert rotational motion into linear motion.
Introduction to robotic kinematics, including joints, linkages, degrees of freedom (DoF), and coordinate systems.
Concepts of motor control hardware, particularly the role of microcontrollers and dedicated driver boards like ODrive.
Advanced Inverse Kinematics (IK) and dynamics to calculate joint trajectories for desired foot placement in 3D space.
Gait design and locomotion algorithms (such as trotting, pacing, and galloping) for coordinating multiple quadruped legs.
Active compliance and force/torque control strategies to allow the leg to absorb impacts and adapt to uneven terrain.
State estimation and sensor fusion, using IMUs and joint encoders to track the robot's orientation and spatial position.
121K views3.3Klikes24:07@jamesbrutonOriginal Release: 2018-06-19

This video demonstrates the prototyping of a single leg for an open-source robot dog (openDog), featuring 12 brushless motors with encoders driving ball screws for precise leg movement. The design uses inverse kinematics to calculate joint angles from desired leg positions, with trigonometric modeling enabling accurate control. Key mechanical components include SFU 16:05 ball screws (16mm diameter, 5mm pitch), 2060 T-slot extrusions for structural support, and V-wheels running in T-slots to prevent rotation. The control system uses ODrive motor drivers with 8,192 encoder counts per revolution, requiring approximately 3,490 encoder clicks for 100mm of linear travel after accounting for belt reduction. The project employs 3D printed prototypes for initial testing before transitioning to CNC-machined aluminum parts, with code published under GPL 3 license on GitHub for community contribution.