Alternatives to Physical Intelligence

Open-source Python SDK and CLI for training and deploying physical AI models — VLA and SLM — from managed cloud compute to real robot hardware.

Skild AI is building an omni-bodied robotic brain intended to control any robot for any physical task.

Rhoda AI builds video-predictive robot foundation models for heavy industrial manipulation, learned from a handful of human demos.

MRPT: open-source C++ libraries and Python bindings for SLAM, 3D vision, and sensor fusion in robotics.

Remoroo teaches industrial robot arms dexterous tasks from one sentence, then proves each skill with an auditable success rate.

Open-source simulated benchmark for language-conditioned, long-horizon robot manipulation from the University of Freiburg.

Trim is a research-stage foundation model that generates how physical systems evolve forward in time.

Zeon Systems turns a plain-language experiment prompt into a robot-executed wet lab protocol, with a world-model preview before anything moves.

License the Nuro Driver, an AI-first L4 self-driving system, for robotaxis, commercial fleets, and personal vehicles.

Open source dev kit for AI-native robotics with leader-follower teleoperation.

Build, code, and program OpenCat quadruped robot kits — Bittle, Nybble Q, and the new Quaddle desk robot dog.

General Instinct compresses VLM and world-action models and runs them on robots and edge silicon with a single inference runtime.

Applied AI research for the physical world: transformer design, autonomous counter-UAS, and dexterous manipulation.

6thSense captures tactile sensing and egocentric touch data for training dexterous robot policies

Twolabs builds post-training environments and datasets for robotics, PCB design, and embedded systems.

Open-source data layer for Physical AI: log, query, transform, visualize, and train on multimodal robotics data in one toolchain.

Wayve AI Driver is mapless, vehicle-agnostic autonomous driving software that scales L2 to L4 autonomy.

Open-core LiDAR SLAM for ROS 2: LiDAR odometry, LIO, 3D SLAM and georeferencing you can modify at the source level.

Open-source autonomous driving simulator with a Python API, ROS2 support, and Unreal Engine 5.5 rendering.

Archetype AI's Newton physical world model turns raw sensor, video, and telemetry data into real-time operational intelligence.

OpenCOOD is an open cooperative perception benchmark for V2V 3D detection and LiDAR fusion research.

Mbodi AI lets your floor staff teach industrial robots new skills by describing the task and demonstrating it — no code, no reprogramming downtime.

Radar sensor for autonomy teams training end-to-end driving policies on raw radar, real or synthetic.

Open-source general-purpose robots that automate unsafe and repetitive tasks.

Build and program an open-source robot dog with a 200 Hz inverse kinematics control loop and a fully hackable hardware stack.

Locus Robotics automates warehouse fulfillment with AI-driven AMRs, robot-to-goods picking, and the LocusONE orchestration platform.

Boston Dynamics Spot is a four-legged mobile inspection robot that climbs stairs, crosses rubble and carries sensor payloads into hazardous industrial sites.

Carbon Robotics pairs the LaserWeeder G2 laser weeding machine with Carbon Autonomy retrofit tractor autonomy for large farms.

Figure AI builds the Figure 03 humanoid robot and the Helix AI stack, deployed at BMW and moving toward home use.

Physical AI infrastructure for automotive, defense, trucking and mining — simulation, Vehicle OS, and self-driving stacks in one platform.