Acknowledged code sources, not a claim of inheriting all weights or capabilities.
A CROSS-DOMAIN READING LENS
From models to the world.
Robotics, autonomous driving and world models: source-grounded connections between general intelligence and physical systems.
Countries and regions follow institutional sources; cross-border work stays visible.
01
From understanding to action
Multimodal knowledge enters robotics while perception and control develop along their own paths.
Qwen · UnifoLM · PointWalker ↗
02From reasoning to driving
How do general models and driving-specific task stacks meet at action?
Gemini · EMMA · Alpamayo ↗
03From generation to worlds
The next frame, a possible future, or an environment for decisions?
GAIA-1 · Cosmos ↗
Source-acknowledged connectionShared methods, not inheritanceEditorial comparison
Understanding, bodies and control
Unitree acknowledges building its project on these two works.
PPO and point-cloud attention meet in a locomotion policy, without a language-dialogue interface.
Reasoning and action on the road
Waymo explicitly builds EMMA on Gemini.
The original R1 architecture pairs a physical-reasoning model with a diffusion trajectory decoder.
An editorial comparison of a specialized task stack and general-model transfer, not inheritance.
Compare language-space representations of robot actions and driving trajectories without merging their evaluations.
Generating futures, testing worlds
Both study conditioned future generation. This is not a code or model inheritance edge.
Research connections are not capability rankings. Simulation, hardware experiments and commercial deployment provide different levels of evidence.
Begin with these works
PointWalker: Mapless Quadruped Locomotion Directly From Raw LiDAR Point Clouds
Fuses LiDAR point clouds and proprioception through attention, connecting terrain perception with a reinforcement-learned quadruped policy.
UnifoLM-WLA-1.0: One Model Driven, Whole-Body Coordination
Connects embodied understanding, interaction-centric world modeling and action learning for tabletop and whole-body humanoid manipulation.
XPeng X-World: Controllable Multi-View Generative World Model for Autonomous Driving
Describes a controllable multi-view world model used for driving simulation, scenario generation, and evaluation.
Kimi K2.5: Visual Agentic Intelligence
Studies joint visual-language training and multi-agent task decomposition in an open multimodal model.
UnifoLM-VLA-0: A Vision-Language-Action (VLA) Framework under UnifoLM Family
Continues training on robot manipulation data, connecting visual-language understanding and spatial instructions to action prediction.
Alpamayo-R1: Bridging Reasoning and Action Prediction for Generalizable Autonomous Driving in the Long Tail
Connects driving reasoning and action prediction through Cosmos-Reason and a diffusion trajectory decoder, studying reasoning-action consistency.
Huawei Qiankun ADS 4: World Engine and World Behavior Model (WEWA)
Huawei's official WEWA description links generated scenarios in a cloud world engine with an on-vehicle world behavior model.
Cosmos World Foundation Model Platform for Physical AI
A platform of video curation, tokenizers, pretrained world models and post-training for physical-AI applications.
EMMA: End-to-End Multimodal Model for Autonomous Driving
Builds on Gemini to predict driving trajectories, perception objects and road structure from camera inputs and text.
SAM 2: Segment Anything in Images and Videos
Extends promptable segmentation from images to video using streaming memory and interactive data collection.
Mixtral of Experts
Introduces a sparse mixture-of-experts model exploring model capacity and activated computation.
GAIA-1: A Generative World Model for Autonomous Driving
Models video, text and vehicle actions as sequences to generate conditioned driving scenes and possible futures.
Planning-oriented Autonomous Driving
Unifies perception, prediction and planning around the driving objective, using query interfaces to coordinate tasks.
The ApolloScape Open Dataset for Autonomous Driving and its Application
Documents a driving dataset and applications for sensor fusion, semantic understanding, and localization.