Skip to content

RoboticsWayve

Wayve

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

Category
Robotics
Pricing
PAID
Hosting
Local
Platforms
API
Models
Self-contained (on-device)
Verified
Jun 9, 2026

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.

Capabilities 4

What it actually does — grouped by capability family.

  • Autonomous navigation (primary capability)
  • Robot foundation model (primary capability)
  • World model (secondary capability)
  • Vision-language-action (secondary capability)

Pros & cons

  • End-to-end learned driving (AV2.0)
  • Hardware-agnostic, any vehicle
  • GAIA generative world model
  • LINGO interpretable driving model
  • Backed by NVIDIA, Microsoft, SoftBank
  • Not a buyable product
  • End-to-end models harder to audit
  • Pre-commercial deployment stage
  • Regulatory path still uncertain

Tags

Further reading

View all Robotics
  • View Genesis details
    RoboticsFREEOSS

    Genesis

    Genesis AI

    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.

    Ultra-fast pure-Python simulation
    Generative data engine still rolling out
    • robotics
    • simulation
    • physics-engine
    • embodied-ai
    • +1
  • View Dyna Robotics details
    RoboticsPAID

    Dyna Robotics

    Dyna Robotics

    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.

    Commercial-grade autonomous dexterity
    Narrow demonstrated task set so far
    • robot-foundation-model
    • embodied-ai
    • dexterity
    • commercial-robotics
  • View Generalist AI details
    RoboticsPAID

    Generalist AI

    Generalist AI

    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.'

    Cross-embodiment foundation model
    Pre-commercial research stage
    • robot-foundation-model
    • embodied-ai
    • dexterity
    • manipulation
  • View LeRobot details
    RoboticsFREEOSS

    LeRobot

    Hugging Face

    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.

    Shared datasets + pretrained VLA policies
    Python/PyTorch skills required
    • robotics
    • open-source
    • pytorch
    • imitation-learning
    • +1
  • View Dexterity details
    RoboticsPAID

    Dexterity

    Dexterity

    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.

    Deployed at FedEx, UPS, GXO today
    Enterprise/industrial only, not buyable direct
    • physical-ai
    • warehouse-robotics
    • manipulation
    • logistics
    • +1
  • View Cobot details
    RoboticsPAID

    Cobot

    Collaborative Robotics

    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.

    Practical wheeled design vs pricey humanoids
    Not a general-purpose humanoid
    • mobile-manipulator
    • warehouse-robotics
    • embodied-ai
    • logistics
  • View Standard Bots details
    RoboticsPAID

    Standard Bots

    Standard Bots

    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.

    Teach it tasks by demonstration
    Significant upfront hardware cost (~$37k+)
    • robotics
    • cobot
    • manufacturing
    • automation
    • +1
  • View FieldAI details
    RoboticsPAID

    FieldAI

    FieldAI

    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.

    Embodiment-agnostic: one brain, many robots
    Enterprise-only, no public pricing
    • robot-foundation-model
    • embodied-ai
    • autonomy
  • View Boston Dynamics details
    RoboticsPAID

    Boston Dynamics

    Boston Dynamics

    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.

    Three robots: Atlas, Spot, Stretch
    No public pricing for Atlas
    • humanoid
    • quadruped
    • embodied-ai
    • industrial
  • View Unitree details
    RoboticsPAID

    Unitree

    Unitree Robotics

    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.

    World's top humanoid seller in 2025
    R1 lacks dexterous hands
    • humanoid
    • quadruped
    • embodied-ai
    • hardware