Petoi | Research
Research Spotlight
Advancing Robotics Education and Innovation
Petoi Bittle robot dog family and Nybble robot cat family - both powered by the open-source OpenCat framework - have been deployed in university robotics labs and featured in academic research worldwide. Studies using Petoi robots cover bio-inspired locomotion and gait control, edge AI and autonomous systems, sim-to-real reinforcement learning, and robotics education accessibility.
Institutions including Carnegie Mellon University, Harvard University, MIT, and research groups around the world have published work built on the Petoi open-source quadruped. Petoi robots have shipped to 60+ countries worldwide.
For research enquiries, lab pricing, and multi-unit institutional orders, contact us.
Bio-Inspired Locomotion & Gaits
Focus on using bio-mimetic neural networks, like Central Pattern Generators (CPGs), and reinforcement learning to generate efficient, natural robotic movements such as walking, bounding, and fall recovery
Edge AI & Autonomous Systems
Center on deploying efficient machine learning (TinyML) & autonomous control software directly onto resource-constrained robots for complex, real-world tasks like semantic navigation & structural health monitoring
Robotics Education & Accessibility
Highlight the use of ultra-low-cost, open-source hardware and global educational networks to make hands-on robotics and machine learning accessible to a wider audience
Petoi robots support C++ & Python API, are compatible with ROS/ROS2, support Raspberry Pi and Nvidia Jetson Nano - making them one of the most capable sub-$400 research platforms for legged robot studies.
Researchers choose Petoi for three reasons:
- cost: Bittle X V2 starts from $319 — a fraction of the cost of prosumer robot dogs like Unitree Go2 Edu(from ~$6,000) or industrial ones like Boston Dynamics' Spot (from ~$75,000).
- openness: all firmware and many STD files available on GitHub under the OpenCat framework
- capability: 9–11 high-performance feedback control servo joints, real quadruped gaits, onboard IMU, and expansion ports for Single-Board Computers (SBCs), edge AI devices, Arduino sensors and modules.
Bio-Inspired Locomotion & Gaits
Quadruped locomotion, how four-legged robots walk, bound, recover from falls, and adapt to terrain - is one of the most active areas of robotics research.
Petoi Bittle X and Nybble Q provide a low-cost, high-DOF platform for testing bio-mimetic control strategies including Central Pattern Generators (CPGs), reinforcement learning-based gait optimization, and event-driven sensorimotor systems.
The following published studies used Petoi hardware to develop and validate locomotion algorithms that would o
Accessibility-Based Clustering for Efficient Learning of Locomotion Skills
Tsinghua University, University of Edinburgh
RoboShape Using Topology Patterns to Scalably
Massachusetts Institute of Technology, Harvard University
Bio-Inspired Locomotion & Gaits
University of Pittsburgh
Symmetry-Guided Reinforcement Learning
Syracuse University
Robot Locomotion Control Using Central Pattern Generator with Non-linear Bio-mimetic Neurons
University of Pittsburgh
Toward autonomous event-based sensorimotor control with supervised gait learning and obstacle avoidance
University of Pittsburgh
Edge AI & Autonomous Systems
Deploying machine learning directly on resource-constrained robots, without cloud connectivity, is the defining challenge of embedded AI robotics.
Researchers have used Petoi Bittle X as a physical testbed for TinyML, sim-to-real transfer, GPT-4 semantic planning, and autonomous structural health monitoring. The platform's ESP32 BiBoard and Raspberry Pi expansion port make it viable for edge inference workloads that typically require larger, more expensive hardware.
Closing the Sim-to-Real Gap for Ultra-Low-Cost, Resource-Constrained, Quadruped Robot Platforms
University of Virginia, Harvard University
CONTROL DE UN ROBOT CUADRÚPEDO DESDE M2OS
Universidad de Cantabria
Mobile structural health monitoring using quadruped robots
Aristotle University of Thessaloniki
Mobile Agent Learning in a Simulation Environment Using Supported Learning and Applications in A Real Environment - Sim-to-real Reinforcement Learning with Nybble Robot Cat
University of Osijek
Semantic Intelligence: Integrating GPT-4 with A* Planning in Low-Cost Robotics
Carnegie Mellon University
Robotics Education & Accessibility
Accessible robotics education - giving students hands-on experience with real legged robots rather than simulations, requires hardware that is affordable, open-source, and genuinely capable.
Petoi Bittle X appears in published research on expanding access to machine learning hardware for underserved student populations, Tiny Robot Learning curriculum design, and global STEM robotics program deployment.
The following studies examine how Petoi and OpenCat enable research-grade robotics education at significantly reduced cost.
Highlights the use of ultra-low-cost, open-source hardware (like the Petoi Bittle) and global educational networks to make hands-on robotics and machine learning accessible to a wider audience
Universitat Politècnica de Catalunya (UPC)
Tiny Robot Learning: Challenges and Directions for Machine Learning in Resource-Constrained Robots
Harvard Univ., CMU, Univ. of Virginia, Google Brain, Delft Univ. of Technology
Tiny Robot Learning: Expanding Access to Edge ML as a Step Towards Accessible Robotics
Barnard College, Columbia University
Citing Petoi in Academic Work
If your research uses Petoi hardware or the OpenCat framework, please use the citation:
Li, R. & Petoi LLC. (2018–present). OpenCat: Open-Source Quadruped Robotic Pet Framework.
Research Enquiries & Institutional Pricing
Petoi supports academic research with institutional pricing, multi-unit lab bundles, technical documentation, and direct developer support. We work with university procurement offices and can supply official quotes for departmental purchase orders.
Get in touch with your institution name, intended use case, and quantity. We respond within 2 business days.
