Robotics AI apps

Computer-vision defect detection for manufacturing, trained from CAD.

Bucket Robotics builds computer-vision defect detection for manufacturing that trains from CAD files and synthetic data instead of hand-labeled photos. It generates simulated defects — burn marks, bumps, breaks — from the CAD that every modern part already has, producing production-ready vision models that deploy in minutes and adapt as parts and lines change. The system integrates into existing production lines without adding new hardware, and has drawn early customers in automotive and defense.

Commercial robot foundation models for autonomous, round-the-clock dexterity.

Dyna Robotics builds general-purpose robots powered by DYNA-1, an embodied AI foundation model aimed at commercial-grade, fully autonomous dexterity. In a public demo, DYNA-1 folded 800+ napkins over a continuous 24-hour run at a reported 99.4% success rate with no human intervention. The Redwood City company targets restaurant, grocery, and other service deployments.

Embodied foundation models that train robots to do real physical work.

Generalist is a frontier AI lab building general-purpose foundation models for the physical world, founded in 2024 by researchers from OpenAI, Google DeepMind, and Boston Dynamics. Its hardware-agnostic GEN-1 model targets dexterous, reliable manipulation across many robot embodiments — factories, warehouses, and homes. The company frames its mission simply: 'we train robots that work.'

A generative, ultra-fast physics platform for robotics and embodied AI.

An open-source simulation platform for general-purpose robotics and embodied-AI learning, rebuilt from the ground up in pure Python. Combines a unified multi-physics engine, a photorealistic renderer, and a generative data engine that turns natural-language prompts into simulation scenes and training data.

Open-source models, datasets, and tools for real-world robotics in PyTorch.

Hugging Face's robotics stack — pretrained policies (ACT, Diffusion Policy, SmolVLA, Pi0), shared datasets, simulation environments, and designs for low-cost manipulator kits. Aims to be the 'Transformers for robotics', lowering the barrier to training and sharing robot-learning models.

Physical AI robots for warehouse manipulation and truck loading.

Dexterity builds 'Physical AI' robots that handle complex manipulation in logistics — including DexR, a truck-loading robot, and Mech, a two-armed mobile 'superhumanoid' that lifts up to 132 lb. Its Foresight world model is trained on over 100 million autonomous actions run in production. Customers include FedEx, UPS, GXO, and Sumitomo.

Proxie, an AI-powered mobile manipulator built for human workplaces.

Cobot (Collaborative Robotics) makes Proxie, a wheeled, two-armed autonomous mobile manipulator designed to move carts, totes, and boxes alongside people rather than replace them. It deliberately avoids costly bipedal humanoid designs, combining omnidirectional mobility with AI-driven situational awareness. Early customers include Maersk, Mayo Clinic, Moderna, and Owens & Minor.

No-code AI robotic arms built for US manufacturing.

Standard Bots makes the Core (formerly RO1), a six-axis collaborative robot arm programmed through no-code software — drag, tap, and teach, with no engineers required. A transformer-based model lets the arm learn intricate tasks by human demonstration, targeting CNC machine tending, palletizing, welding, and pick-and-place. The arms are designed and built in the US.

One autonomy brain for all robots.

FieldAI builds embodiment-agnostic foundation models that act as a general-purpose 'brain' for robots, letting a single autonomy stack drive many different robot bodies. Its risk-aware Field Foundation Models use a Belief World Model to operate in dynamic, unmapped environments without GPS or predefined trajectories. The company targets construction, industrial, energy, mining, and other real-world deployments.

Atlas humanoid, Spot quadruped, and Stretch logistics robots from the dynamics pioneer.

The Hyundai-owned robotics company behind the Atlas humanoid, the Spot quadruped, and the Stretch warehouse robot. The production electric Atlas debuted at CES 2026, with its first fleets committed to Hyundai's Metaplant and Google DeepMind and factory deployment in Savannah targeted by 2028. Spot and Stretch are already sold commercially for industrial inspection and logistics.

Quadruped and humanoid robots you can buy at consumer prices.

Unitree Robotics builds quadruped robots (Go2, B2), humanoid robots (G1, H1, R1), robotic arms, and LiDAR systems you can actually order today. It shipped more than 5,500 humanoids in 2025 — roughly a third of global sales — and turned its first profit the same year, rare in a field of pre-revenue rivals.

Apollo, a general-purpose humanoid robot for logistics and manufacturing.

Austin-based Apptronik builds Apollo, a 5'8", 160 lb general-purpose humanoid robot with a 55 lb payload and swappable four-hour battery packs, designed to work in spaces built for humans. Apollo is being piloted in factories and warehouses with Mercedes-Benz and GXO Logistics under a robot-as-a-service model, and a strategic partnership with Google DeepMind pairs the hardware with Gemini Robotics models.

Maker of Digit, the humanoid robot already doing paid work in warehouses.

Agility Robotics builds Digit, a bipedal humanoid for warehouse and factory material handling, orchestrated by its Arc cloud platform. Digit works production shifts at GXO — 100,000+ totes moved at a single site — with commercial agreements at Schaeffler, Mercado Libre, and Toyota Motor Manufacturing Canada. Robots are assembled at the company's RoboFab factory in Salem, Oregon.

Phoenix, a general-purpose humanoid robot driven by the Carbon AI control system.

Sanctuary AI builds Phoenix, a general-purpose humanoid robot for work, powered by Carbon, a hybrid cognitive control system that blends symbolic reasoning, large language models, and reinforcement learning to mimic human cognitive subsystems. Phoenix features highly dexterous hands and can be taught new tasks in under 24 hours.

Humanoid robots powered by Helix, an in-house vision-language-action model.

Figure AI builds general-purpose humanoid robots (Figure 01–03) driven by Helix, its proprietary vision-language-action model that fuses perception, language, and learned control. Figure 03, announced in 2025, targets the home and broader applications; earlier models were piloted in BMW's automotive manufacturing.

A single, omni-bodied foundation model designed to control any robot.

Skild AI builds Skild Brain, a general-purpose robotics foundation model intended to control diverse embodiments — quadrupeds, humanoids, tabletop arms, mobile manipulators — without prior knowledge of a robot's exact form. It pretrains on large-scale simulation and internet video, then post-trains on targeted real-world data for customer deployments.

NEO, a home humanoid robot driven by the onboard Redwood vision-language-action model.

1X (1X Technologies) builds NEO, a household humanoid robot powered by Redwood, a 160M-parameter vision-language-action model that jointly controls locomotion and manipulation for tasks like laundry and answering the door. Redwood runs on NEO's onboard GPU and pairs with an off-board language model for voice control. NEO ships from 2026 at $20,000 or $499/month.

General-purpose foundation models that aim to control any robot to do any task.

Physical Intelligence is a San Francisco lab building general-purpose AI for the physical world. Its π0 (pi-zero) vision-language-action model, built on a pretrained VLM and trained across many robot embodiments, can perform dexterous tasks like folding laundry. The company open-sourced π0's code and weights via its openpi repository.

End-to-end embodied AI for self-driving that scales across any vehicle.

Wayve builds AV2.0, an embodied-AI approach to automated driving that replaces hand-engineered perception-planning-control pipelines with a single end-to-end neural network mapping raw camera and radar input to driving commands. It develops the GAIA generative world model for training and the LINGO language model for interpretable driving. Backed by NVIDIA, Microsoft, and SoftBank, and partnered with Uber.