Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

LMDrive Closed-Loop End-to-End Driving with LLM An end-to-end, closed-loop, language-based autonomous driving framework, which interacts with the dynamic environment via multi-modal multi-view sensor data and natural language instructions

rsasaki0109

7,187 subscribers

10,085 views • 2 years ago •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

5 Comments

wwwwg2 years ago

@Memdotai mem it

Kirito (e/acc) 🏴‍☠️2 years ago

@yassineyousfi_

James2 years ago

Realtime?

toshi_4562 years ago

Does Notice in the demo video refer to Notice Instruction?

Mc Laurin2 years ago

@CaesarFdez :P

Related Videos

Physical AI just got a major data upgrade. NATIX joins The Autoware Foundation 🚗🚌🏎️ as a premium member 🌟 We will provide global multi-camera driving data for the open-source end-to-end autonomous driving model.

Physical AI just got a major data upgrade. NATIX joins The Autoware Foundation 🚗🚌🏎️ as a premium member 🌟 We will provide global multi-camera driving data for the open-source end-to-end autonomous driving model.

NATIX Network

10,772 views • 2 months ago

LEGO-SLAM: Language-Embedded Gaussian Optimization SLAM LEGO-SLAM running at 15 FPS on a ScanNet scene with language-based loop closing for drift correction. LEGO-SLAM is a 3DGS-based SLAM framework that supports open-vocabulary semantic querying and rendering. It tracks via G-ICP and efficiently builds a map by embedding Gaussians with scene-adaptive 16D language features. Map management is achieved through Language Pruning and Language-Based Loop Detection. The generated map enables open-vocabulary 3D Object Localization.

LEGO-SLAM: Language-Embedded Gaussian Optimization SLAM LEGO-SLAM running at 15 FPS on a ScanNet scene with language-based loop closing for drift correction. LEGO-SLAM is a 3DGS-based SLAM framework that supports open-vocabulary semantic querying and rendering. It tracks via G-ICP and efficiently builds a map by embedding Gaussians with scene-adaptive 16D language features. Map management is achieved through Language Pruning and Language-Based Loop Detection. The generated map enables open-vocabulary 3D Object Localization.

Ryohei Sasaki@engineer

14,935 views • 2 months ago

What happens when you introduce natural language to autonomous driving? 👋 Say hello to #LINGO1 Wayve’s open-loop driving commentator 🗣🚘 💬 #EmbodiedIntelligence #AutonomousVehicles #AI

What happens when you introduce natural language to autonomous driving? 👋 Say hello to #LINGO1 Wayve’s open-loop driving commentator 🗣🚘 💬 #EmbodiedIntelligence #AutonomousVehicles #AI

Wayve

37,163 views • 2 years ago

DiffusionDrive Truncated Diffusion Model for End-to-End Autonomous Driving

DiffusionDrive Truncated Diffusion Model for End-to-End Autonomous Driving

AK

17,797 views • 1 year ago

Hyundai Motor's self-driving group '42Dot' has released an 'End-to-End' self-driving video. 42Dot introduces images of the Atria AI experiment drive, adding the following explanation. - Atria AI is an end-to-end driving-based autonomous driving technology developed independently by 42dot that recognizes the surroundings and drives with only eight cameras on the vehicle without relying on an HD map. - It also uses the 400 TOPS class NPU to implement Navigation on Autopilot (NOA) capabilities on highways and cities, and is a prime example of the direction of software-defined autonomous driving technology. What do you think?

Hyundai Motor's self-driving group '42Dot' has released an 'End-to-End' self-driving video. 42Dot introduces images of the Atria AI experiment drive, adding the following explanation. - Atria AI is an end-to-end driving-based autonomous driving technology developed independently by 42dot that recognizes the surroundings and drives with only eight cameras on the vehicle without relying on an HD map. - It also uses the 400 TOPS class NPU to implement Navigation on Autopilot (NOA) capabilities on highways and cities, and is a prime example of the direction of software-defined autonomous driving technology. What do you think?

Tsla Chan

90,628 views • 6 months ago

This is the COMPLETE ~1.5 mile route newly produced autonomous Tesla Model Ys take from production line, to the End-Of-Line Facility. This isn’t a closed loop. Thousands of autonomous vehicles share the road with pedestrians, traffic, and bikers.

This is the COMPLETE ~1.5 mile route newly produced autonomous Tesla Model Ys take from production line, to the End-Of-Line Facility. This isn’t a closed loop. Thousands of autonomous vehicles share the road with pedestrians, traffic, and bikers.

Nic Cruz Patane

32,271 views • 1 year ago

end-to-end self driving is the way

end-to-end self driving is the way

ΛI DRIVR

66,446 views • 11 months ago

🚨NEWS: HYUNDAI MOTOR GROUP TO ADOPT $TSLA-LIKE FSD APPROACH TO ACCELERATE AUTONOMOUS DRIVING • Hyundai Motor Group is actively incorporating an “End-to-End (E2E)” autonomous driving approach similar to Tesla’s FSD (Full Self-Driving). • The company plans to combine this data-driven E2E method with its existing LiDAR-based technology (Motional) for greater synergy. ✅ Key Strategy • Presented E2E autonomous driving technology internalization as a new direction at the Kia CEO Investor Day. • Direction set after the appointment of Park Min-woo (former Tesla AVP head) as CEO of 42dot. ✅ Development Goals • Level 2+ on highways by the end of 2027 • Level 2++ capable of driving in urban environments by 2029 ✅ Organizational Expansion • 42dot workforce increased 38.6% year-over-year to 876 employees as of end of last year. • Hyundai Autoever is responsible for developing digital twin-based autonomous driving maps.

🚨NEWS: HYUNDAI MOTOR GROUP TO ADOPT $TSLA-LIKE FSD APPROACH TO ACCELERATE AUTONOMOUS DRIVING • Hyundai Motor Group is actively incorporating an “End-to-End (E2E)” autonomous driving approach similar to Tesla’s FSD (Full Self-Driving). • The company plans to combine this data-driven E2E method with its existing LiDAR-based technology (Motional) for greater synergy. ✅ Key Strategy • Presented E2E autonomous driving technology internalization as a new direction at the Kia CEO Investor Day. • Direction set after the appointment of Park Min-woo (former Tesla AVP head) as CEO of 42dot. ✅ Development Goals • Level 2+ on highways by the end of 2027 • Level 2++ capable of driving in urban environments by 2029 ✅ Organizational Expansion • 42dot workforce increased 38.6% year-over-year to 876 employees as of end of last year. • Hyundai Autoever is responsible for developing digital twin-based autonomous driving maps.

Tsla Archive

23,202 views • 2 months ago

"NVIDIA’s Jensen Huang confirms that Tesla stack is the most advanced AV stack globally. With a focus on end-to-end AI and unmatched operations, Tesla continues to set the bar for autonomous driving

"NVIDIA’s Jensen Huang confirms that Tesla stack is the most advanced AV stack globally. With a focus on end-to-end AI and unmatched operations, Tesla continues to set the bar for autonomous driving

X Freeze

686,532 views • 3 months ago

Command humanoids *directly* with natural language? Introducing LangWBC, a generative, end-to-end policy that turns natural language into real-world whole-body humanoid control! 💬→🦿Smooth, robust, surprisingly intuitive! See more 👉 #RSS2025

Command humanoids directly with natural language? Introducing LangWBC, a generative, end-to-end policy that turns natural language into real-world whole-body humanoid control! 💬→🦿Smooth, robust, surprisingly intuitive! See more 👉 #RSS2025

Zhongyu Li

14,142 views • 1 year ago

We got a fully open source, end-to-end, conversational AI that you can run on a MacBook Air before Multi-modal GPT4o!

We got a fully open source, end-to-end, conversational AI that you can run on a MacBook Air before Multi-modal GPT4o!

Vaibhav (VB) Srivastav

209,098 views • 1 year ago

Anthropic CEO, Dario Amodei: "we might be 6-12 months away from models doing all of what software engineers do end-to-end" We're approaching a feedback loop where AI builds better AI But the loop isn't fully closed yet, chip manufacturing and training time still limit speed

Anthropic CEO, Dario Amodei: "we might be 6-12 months away from models doing all of what software engineers do end-to-end" We're approaching a feedback loop where AI builds better AI But the loop isn't fully closed yet, chip manufacturing and training time still limit speed

Haider.

743,313 views • 4 months ago

The next frontier of autonomous driving is unlocked by reasoning models. NVIDIA Alpamayo brings together open AI models with reasoning capabilities, closed-loop simulation tools, and massive real-world driving datasets. Alpamayo 1 is a vision–language–action model that explains its own decisions through explicit reasoning traces, enabling trustworthy, humanlike decision-making. Together with NVIDIA’s Physical AI dataset and AlpaSim simulation, Alpamayo provides the tools and scale required to enable level 4 autonomous vehicles. ▶️ Watch now:

The next frontier of autonomous driving is unlocked by reasoning models. NVIDIA Alpamayo brings together open AI models with reasoning capabilities, closed-loop simulation tools, and massive real-world driving datasets. Alpamayo 1 is a vision–language–action model that explains its own decisions through explicit reasoning traces, enabling trustworthy, humanlike decision-making. Together with NVIDIA’s Physical AI dataset and AlpaSim simulation, Alpamayo provides the tools and scale required to enable level 4 autonomous vehicles. ▶️ Watch now:

NVIDIA DRIVE

35,324 views • 5 months ago

🎤🎤 Excited to introduce COME-robot🤖🤖, Closed-Loop Open-Vocabulary Mobile Manipulation with GPT-4V. It is the first closed-loop framework utilizing the vision-language foundation model for open-ended reasoning and adaptive planning in real-world scenarios. COME-robot demonstrates a significant improvement in task success rate (~25%) compared to SOTA methods. Project: Arxiv:

🎤🎤 Excited to introduce COME-robot🤖🤖, Closed-Loop Open-Vocabulary Mobile Manipulation with GPT-4V. It is the first closed-loop framework utilizing the vision-language foundation model for open-ended reasoning and adaptive planning in real-world scenarios. COME-robot demonstrates a significant improvement in task success rate (~25%) compared to SOTA methods. Project: Arxiv:

Siyuan Huang

22,291 views • 2 years ago

🚀Our first AI NFT Agent showcase! 🚀 Featuring jAIhoz.ron , Kojin Genesis Agent 406 (Kenji) and 267 (Haruto). Watch till the end to see the multi-language features.

🚀Our first AI NFT Agent showcase! 🚀 Featuring jAIhoz.ron , Kojin Genesis Agent 406 (Kenji) and 267 (Haruto). Watch till the end to see the multi-language features.

Kojin

97,192 views • 1 year ago

The smoothest most natural self-driving I've ever experienced was built entirely around computer vision, with end to end neural nets. It is clearly the way forward. Watch in 4K: Raw 1x footage:

The smoothest most natural self-driving I've ever experienced was built entirely around computer vision, with end to end neural nets. It is clearly the way forward. Watch in 4K: Raw 1x footage:

Whole Mars Catalog

666,476 views • 1 year ago

Autonomous end to end whole body mobile manipulation, via Noble Machines. All one policy.

Autonomous end to end whole body mobile manipulation, via Noble Machines. All one policy.

Chris Paxton

12,023 views • 3 months ago

TimeFusion is built on a new multimodal encoding and decoding framework based on Universal Tokens. This approach can easily extend to a wide range of modalities beyond time-series data and language. Here’s a preview of our ongoing work using Universal Tokens to unify images, sensor data, and language in a single embedding space. This allows dynamic cross-modal mappings—such as converting a video stream directly into physical measurements, all controlled through natural language. #PhysicalAI #multimodal

TimeFusion is built on a new multimodal encoding and decoding framework based on Universal Tokens. This approach can easily extend to a wide range of modalities beyond time-series data and language. Here’s a preview of our ongoing work using Universal Tokens to unify images, sensor data, and language in a single embedding space. This allows dynamic cross-modal mappings—such as converting a video stream directly into physical measurements, all controlled through natural language. #PhysicalAI #multimodal

Ivan Poupyrev

36,361 views • 5 months ago

Su Qing, Vice President and Chief Architect at Horizon — and former head of intelligent driving products at Huawei — said that Tesla’s FSD is generationally ahead of its Chinese competitors in certain respects. He noted that domestic systems, including Horizon’s, are in fact hybrid architectures combining end-to-end models with rule-based components, even though they are marketed as fully end-to-end solutions. According to Su, the key constraints facing Chinese autonomous driving systems — relative to FSD — include limited large-scale training compute, frequent and unpredictable infrastructure changes, and highly irregular driving behaviors. These uniquely complex road conditions have led Chinese developers to adopt a more programmatic, hybrid approach to autonomy rather than a purely end-to-end system. He said the FSD team’s level of expertise and spirit of innovation both deserve real admiration.

Su Qing, Vice President and Chief Architect at Horizon — and former head of intelligent driving products at Huawei — said that Tesla’s FSD is generationally ahead of its Chinese competitors in certain respects. He noted that domestic systems, including Horizon’s, are in fact hybrid architectures combining end-to-end models with rule-based components, even though they are marketed as fully end-to-end solutions. According to Su, the key constraints facing Chinese autonomous driving systems — relative to FSD — include limited large-scale training compute, frequent and unpredictable infrastructure changes, and highly irregular driving behaviors. These uniquely complex road conditions have led Chinese developers to adopt a more programmatic, hybrid approach to autonomy rather than a purely end-to-end system. He said the FSD team’s level of expertise and spirit of innovation both deserve real admiration.

Ray

83,127 views • 4 months ago

a priest blessed one end zone pregame… the same end zone where Loop shanked the FG

a priest blessed one end zone pregame… the same end zone where Loop shanked the FG

Warren Sharp

199,807 views • 5 months ago