Bipedal Humanoid Robotic System Development

Request for Proposals (RFP) / Freelance Job Specification: Custom Bipedal Humanoid Platform
Role Type: Contract / Freelance (Full-system or Module-specific)
Location: Remote (Hardware prototyping local or shipped)
Engagement: Milestone-based contract
1. Project Overview
We are designing and fabricating a custom open-architecture bipedal humanoid robot. The goal is to build a functional, anthropomorphic platform capable of dynamic balance, stable bipedal locomotion, and compliant interaction.
Because building a complete humanoid spans multiple disciplines, we are hiring specialized engineers and domain experts. Applicants may bid on the full lifecycle or submit proposals for individual tracks aligned with their core expertise.
2. Specialized Tracks & Scope of Work
Track A: Mechanical Engineering & CAD Design
Primary Focus: Kinematic layout, structural integrity, thermal dissipation, and manufacturability.
Kinematic Architecture: Design 20+ to 30 Degrees of Freedom (DoF) across legs, waist/torso, arms, and neck with anthropomorphic mass distribution.
CAD Modeling: Full 3D assembly models in Onshape, SolidWorks, or Fusion 360, including tolerances, fastener callouts, bearing selections, and cable-routing channels.
Actuator Selection & Mechanical Integration: Sizing and mounting of Quasi-Direct Drive (QDD) or planetary/cycloidal brushless DC (BLDC) actuators.
Design for Manufacturing (DFM):
Structural load-bearing members optimized for CNC 6061/7075 aluminum or carbon fiber.
Rapid-prototyping components optimized for additive manufacturing (MJF, SLS Nylon, or PETG/CF-PLA).
Deliverables: Complete STEP/Parasolid assembly files, 2D production drawings with GD&T, unified hardware Bill of Materials (BOM), and mechanical assembly documentation.
Track B: Electrical Systems, Power Architecture & Gate Drivers
Primary Focus: High-current power distribution, safety isolation, and switching electronics.
Power Distribution System (PDS):
Multi-rail power distribution board (PDB) design supporting 24V–48V Li-ion/LiFePO4 battery architectures.
Regulated DC-DC step-downs (5V, 12V, 19V/24V) for onboard computing, microcontrollers, and sensors.
Integrated safety: soft-start circuits, emergency-stop (E-stop) interlocks, reverse polarity protection, and transient voltage suppression (TVS).
Power Electronics & Gate Driving:
Design or specification of high-frequency 3-phase inverter stages and half-bridge/full-bridge gate driver circuits for BLDC commutation.
Low-side and high-side MOSFET/GaN FET driver selection with shoot-through protection and dead-time management.
Harness & Signal Integrity:
Wiring diagrams, harness routings, connector selection (Molex, XT90, JST, Binder), and shielding against electromagnetic interference (EMI).
Deliverables: Altium/KiCad schematic files, PCB gerbers, layout files, electrical wiring diagrams, harness pinout charts, and power budget calculations.
Track C: Embedded Systems, Low-Level Commutation & Gate Logic
Primary Focus: Real-time motor control, digital logic, and communication buses.
Actuator Firmware & Motor Control:
Implementation of Field Oriented Control (FOC), Space Vector PWM (SVPWM), and high-frequency current sensing loops (≥20–40 kHz).
Position, velocity, and torque control with magnetic absolute encoder integration (BiSS-C, SPI, SSI).
FPGA / Digital Gate Logic (if applicable):
Verilog/VHDL logic design for multi-channel synchronous motor commutation, hardware-timed encoder decoders, and deterministic fault monitoring.
Communication Bus:
High-speed communication protocol implementation (CAN FD, EtherCAT, or RS-485) between the central controller and distributed joint drivers (≥1 kHz loop rate).
Deliverables: Firmware source code, FPGA bitstreams/logic specifications, register maps, communication interfaces, and low-level debugging utilities.
Track D: Gait Generation, Dynamic Balancing & Locomotion
Primary Focus: Physics simulation, control theory, and Sim-to-Real deployment.
Simulation Setup:
Rigorous URDF/MJCF kinematic and dynamic model generation with precise inertia matrices, mass distribution, and friction parameters.
Setup in Isaac Lab / Isaac Gym or MuJoCo.
Locomotion Policy / Control Architecture:
Option 1 (Reinforcement Learning): Train robust, blind/perceptive bipedal locomotion policies using domain randomization (terrain, friction, latency, external disturbances).
Option 2 (Classical Control): Implement Model Predictive Control (MPC), Whole-Body Control (WBC), and Virtual Model Control for balance recovery and push-resistance.
Sim-to-Real Transfer:
Export policies (ONNX / TensorRT / TorchScript) and deploy them to embedded real-time microprocessors with latency validation under 5 ms.
Deliverables: Training pipelines, simulation repository, exported gait models, deployment inference runtime, and empirical tuning guidelines.
Track E: Systems Integration, Compute & Middleware
Primary Focus: Real-time OS, central compute orchestration, and high-level interfaces.
Central Compute Architecture:
Configuration of onboard computing (e.g., Jetson Orin Nano/AGX, Radxa CM5, or x86 SBC) running real-time Linux (⁠PREEMPT_RT⁠).
Middleware Integration:
ROS 2 (Humble/Jazzy) or custom zero-copy middleware (e.g., Zenoh, CycloneDDS) connecting state estimation to low-level motor drivers.
State Estimation:
Extended Kalman Filter (EKF) combining high-rate IMU data, joint kinematics, and contact sensors to compute floating-base pose and velocity.
Deliverables: Operating system images/Docker environments, ROS 2 workspace packages, state estimation nodes, and telemetry/diagnostic dashboards.
3. Intellectual Property (IP) & Commercial Terms
Work-for-Hire: All designs, schematics, source code, CAD files, and models developed under this engagement will be exclusive property of the client upon milestone payment.
Licensing Compliance: No copyleft viral licenses (e.g., GPL, AGPL, CERN-OHL-S) may be introduced into proprietary deliverables without written pre-approval. Permissive open-source libraries (MIT, Apache 2.0, BSD) are welcome where appropriate.
Non-Disclosure: A mutual NDA will be executed prior to sharing detailed project assets.
4. How to Apply
Please submit your proposal specifying your track(s) of interest along with:
1. Applied Track(s): (e.g., "Track A: Mechanical" or "Track D: Gait & Locomotion")
2. Relevant Portfolio: Links to GitHub repositories, GrabCAD/CAD portfolios, videos of physical walking robots, custom motor controllers, or hardware builds you have personally engineered.
3. Proposed Tech Stack: Summary of your preferred tools for this engagement (e.g., CAD software, simulation framework, motor drivers, compute boards).
4. Estimated Timeline & Milestone Rates: Hourly rate or fixed milestone cost estimate for your designated track.