Toward Next-Gen Physical AI: Research Group at Applied Intuition

문의하기

도전

ADAS 및 AD는 데이터, 머신러닝(ML), 시뮬레이션에 대한 새로운 접근 방식이 필요한 엔드투엔드(E2E) 아키텍처를 점점 더 많이 활용하고 있습니다. 차량 내 경험은 인포테인먼트, 모바일 애플리케이션, 차량 제어 시스템에 이르기까지 소프트웨어 정의 방식으로 전환되고 있습니다. 자동차 제조사들은 UN R157 및 UN R171과 같은 규정을 준수하며 개발 속도를 높이고 차세대 기능을 개발해야 합니다.

This group, led by Dr. Wei Zhan, includes leading experts from top institutions and companies, recognized for their exceptional academic and industry contributions—including eight Best Paper awards at premier conferences and journals such as CVPR and ICRA. Join our group to work with a team of top researchers who each have thousands of citations across their work and many top-tier publications, including a CVPR Best Paper lead author, to advance next-generation physical AI.

제품 살펴보기

Massive data from large fleets processed by automated data engine

AI 기반 시뮬레이터로 운전 로그를 상호작용 가능한 세계로 변환하여 폐쇄 루프 ADAS 테스트 및 훈련을 수행합니다

Various fleets to test and deploy our autonomy — including cars, trucks, and offroad mining

Large fleets across various products globally to test and deploy autonomy algorithms developed by our researchers and engineers, from autonomous cars tested in challenging urban scenarios to self-driving trucks and mining/construction vehicles for both onroad and offroad scenes.

Robot fleet and human data collection

Robot fleets with humanoids, mobile manipulators, and table-top dual-arm robots with cutting-edge, tactile-aware dexterous hands, as well as various human data collection devices such as MoCap, headset, and gloves.

Large-scale ML infrastructure and neural simulation with efficient tools

As a world-leading tool provider for autonomous systems, our researchers are supported by various efficient tools, high-quality neural simulation and synthetic data at scale in closed loop, and ML infrastructure toward large-scale training with thousands or more of GPUs.

Research Focuses

We address the following research fields in our group to advance next-generation physical AI:

World-action foundation model pre-training

The next-generation foundation model for physical AI should be pretrained with better balanced and grounded multi-modal data from ego action to world (vision, behavior, physics) and language toward the concrete tasks. We accordingly address the following topics:
Feed-forward/generative 4D reconstruction and world foundation model for reactive generation of 4D world conditioned on ego action with high throughput
Pretraining of world-action model with grounded modalities including vision, physics, and language

Reinforcement learning and foundation model post-training

Post-training of foundation models such as world-action and vision-language-action models is highly crucial for performance improvement, safety assurance and preference alignment for physical AI applications. In view of this, we address the following research fields:
Closed-loop reinforcement-learning-based post-training of foundation models supported by high-fidelity, scalable, high-throughput simulation constructed/learned from large-scale real-world data
Self-play reinforcement learning supported by high-throughput simulation in combination with human data imitation toward robust and human-like physical AI even with low data.

Robot learning and data

Toward capable robotic generalists, physics-aware modalities become demanding and the large-scale data to enable the generalists is not as straightforward to obtain as driving applications. Therefore, we have special focus on various paradigms and methods of robot learning and data to properly scale up data from robot, human, and synthetic world with physics-aware modalities, and corresponding design methods for utilization of the data to achieve robotic generalists.

Publications and Projects

S2GO: Streaming Sparse Gaussian Occupancy Prediction

Jinhyung Park, Chensheng Peng, Yihan Hu, Wenzhao Zheng, Kris Kitani, Wei Zhan
ICLR 2026
arXiv Preprint
Website

SPACeR: Self-Play Anchoring with Centralized Reference Models

Wei-Jer Chang, Akshay Rangesh, Kevin Joseph, Matthew Strong, Masayoshi Tomizuka, Yihan Hu, Wei Zhan
ICLR 2026
arXiv Preprint
Website

Open Positions

Self-Driving

Research Scientist - 3D Vision and Generation, Self-Driving
Research Scientist - Reinforcement Learning, Self-Driving
Research Engineer - 3D Vision and Generation, Self-Driving
Research Engineer - Reinforcement Learning, Self-Driving
Research Intern - 3D Vision and Generation, Self-Driving
Research Intern - Reinforcement Learning, Self-Driving

Robotics

Research Scientist - World-Action Foundation Model, Robotics
Research Scientist - Reinforcement Learning, Robotics
Research Engineer - Robot Learning
Research Engineer - Robotic Hardware, Simulation and Data
Research Intern - World-Action Foundation Model, Robotics
Research Intern - Reinforcement Learning, Robotics
Research Intern - Robotic Hardware, Simulation and Data

ML infrastructure

Research Engineer - AI/RL Infrastructure
Applied Intuition의 미션은 안전하고 지능화된 기계의 전 세계 도입 가속화입니다.
채용 중입니다!
Applied Intuition의 채용 정보에 대해 알아보고 지원하세요.
채용 정보 보기
en
ja
ko
차량 OS 제품
온보드 플랫폼오프보드 플랫폼워크벤치하드웨어캐빈 인텔리전스
자율주행 시스템 (SDS)
차량용 SDS광산 및 건설용 SDS트럭용 SDS
시뮬레이션 제품
Object SimSensor SimLog SimCloud EngineVehicleSimHIL SimNeural Sim
데이터 제품
Data ExplorerSynthetic DatasetsValidation ToolsetTest SuitesMap ToolsetApplied Intuition Copilot
솔루션
차량 OS자율주행 시스템 (SDS)차량 인텔리전스툴
산업 분야
자동차트럭국방 ↗광산건설농업
고객
ToyotaNissanPorsche AGKomatsuIsuzuTorcVolkswagen GroupValeoKodiakMay MobilityNational InstrumentsStellantisScientific SystemsAISIN
회사
블로그이벤트채용 정보소개뉴스룸개인정보 처리방침연락처Research
© 2024 Applied Intuition, Inc.
Applied Intuition Defense ↗