Physicalintelligence

Controls Engineer

6mo ago
USA
Physicalintelligence

Controls Engineer

6mo ago
USApidlqrmpcinverse dynamicsreal-time loopsneural networkscanspi+2

Design and implement control algorithms and real-time loops to ensure stable and reliable robot behavior.

Responsibilities

  • Design & implement control algorithms: PID, LQR, MPC, inverse dynamics, and feedforward controllers.
  • Build & validate models: Create and refine physical and inverse dynamics models for simulation and control design.
  • Develop real-time loops: Write and optimize runtime control loops, including neural-network-driven control.
  • Own robotic bring-up: Integrate and tune arms, mobile bases, teleop systems, and full-body platforms.
  • Debug complex system behaviors: Diagnose and resolve hardware/software/runtime issues using first-principles reasoning.
  • Build sensor/actuator subsystems: Work with embedded systems, drivers, and communication protocols (CAN, SPI, I2C, Ethernet).
  • Partner cross-functionally: Work with researchers, platform engineers, and operators to ensure stable, predictable real-world behavior.
  • Support R&D: Prototype configurations, collect structured datasets, and iterate directly with researchers.

Nice to have

  • Background in manipulation or mobile robotic platforms
  • Exposure to robot learning or integrating learned policies into control stacks
  • Ability to design or refine custom actuator or sensor hardware
  • Pursuant to the San Francisco Fair Chance Ordinance, we will consider for employment qualified applicants with arrest and conviction records.

Other

  • As a Controls Engineer, you will design and implement the algorithms that make PI’s robots behave predictably, smoothly, and safely under varied and uncertain conditions.
  • The Controls team builds and tunes the core feedback and model-based algorithms, real-time loops, simulations, and actuator/sensor subsystems that make PI’s robots stable and reliable. They work closely with research, hardware, and operations to debug complex system behaviors and ensure our learning-based systems operate under strict real-time constraints in unpredictable environments.
  • Deep understanding of model-based control algorithms and inverse dynamics
  • Ability to validate control approaches in simulation and translate them to real hardware
  • Proficiency in Python and C++, including firmware-adjacent development
  • Skill in writing and tuning real-time control loops
  • Hands-on capability to debug electromechanical systems end-to-end
  • Familiarity with embedded communication protocols (CAN, SPI, I2C, Ethernet)
  • Clear communication with researchers, hardware teams, and operators
  • A structured, collaborative approach to solving complex system issues