DeepRoute.ai (@DeepRoute_ai) header
DeepRoute.ai (@DeepRoute_ai) profile picture

Founded in 2019, DeepRoute is dedicated to the research, development and application of autonomous driving technology, with a mission to create AGI in robots.

San Francisco Bay Area, CA
Joined August 2019
Photo by DeepRoute.ai (@DeepRoute_ai)Photo by DeepRoute.ai (@DeepRoute_ai)Photo by DeepRoute.ai (@DeepRoute_ai)Photo by DeepRoute.ai (@DeepRoute_ai)Photo by DeepRoute.ai (@DeepRoute_ai)Photo by DeepRoute.ai (@DeepRoute_ai)Photo by DeepRoute.ai (@DeepRoute_ai)Photo by DeepRoute.ai (@DeepRoute_ai)Photo by DeepRoute.ai (@DeepRoute_ai)Photo by DeepRoute.ai (@DeepRoute_ai)
#APEC2026. Shenzhen. Physical AI in motion. ๐ŸŒ๐Ÿš˜ As an official APEC 2026 corporate visit site, we welcomed international guests and media to experience our assisted driving and Robotaxi technologies firsthand. From Shenzhen to the world. ๐ŸŽฌ #PhysicalAI #Robotaxi
2
2
88
Yesterday: our #APEC2026 showcase. Today: the tech behind it. We're scaling foundation models and exploring RSI at Superfluid Lab to tackle complex driving scenarios. One metric we're aiming to improve: MPCI, the distance between critical driver interventions. #PhysicalAI
2
103
Physical AI, meet the real world. ๐Ÿš˜ DeepRoute.ai is a designated #APEC2026 company visit site in Shenzhen, showcasing foundation model research, assisted driving and Robotaxi. 600,000+ vehicles are already equipped with our assisted driving solutions. #APEC2026
1
4
392
200 million miles. Still more homework. At ECCV, we talked world models, self-play, and why even the smartest AI drivers still have a lot to learn. Turns out, the long tail never really ends. Our notes ๐Ÿ‘‡ ๐Ÿ‡ฌ๐Ÿ‡งlinkedin.com/feed/update/urnโ€ฆ ๐Ÿ‡ฐ๐Ÿ‡ท linkedin.com/feed/update/urnโ€ฆ
1
3
195
์ •์ผ์šฐ์™€ ํ•จ๊ป˜ํ•˜๋Š” ์ƒํ•˜์ด ๋“œ๋ผ์ด๋ธŒ ๐Ÿš— DeepRoute.ai์˜ ์ง€๋Šฅํ˜• ์ฃผํ–‰ ๋ณด์กฐ ๊ธฐ์ˆ ์„ ๊ฒฝํ—˜ํ•œ ๋ฐฐ์šฐ ์ •์ผ์šฐ. ํƒ‘์Šน ํ›„์˜ ๊ถ๊ธˆ์ฆ๋ถ€ํ„ฐ ๋„์ฐฉ ํ›„์˜ ์†Œ๊ฐ๊นŒ์ง€, ์˜์ƒ์œผ๋กœ ๋งŒ๋‚˜๋ณด์„ธ์š”. #์ •์ผ์šฐ #DeepRouteAI
1
6
308
AI driving intelligence. Built for production vehicles.๐Ÿš— Revisiting Chosun Ilboโ€™s June 19, 2026 interview with DeepRoute.ai co-founder Wang Tian. ์กฐ์„ ์ผ๋ณด๊ฐ€ ๋งŒ๋‚œ DeepRoute.ai. Watch the interview: youtube.com/watch?v=nbB3juV8โ€ฆ #DeepRouteAI #AutonomousDriving
7
495
Physical AI isn't a concept. It's what we ship. โ€ข 300K vehicles deployed โ€ข 1.3B km real-world data โ€ข Near hourly system updates Every km makes us stronger. Next stop: million-scale by 2026. ๐Ÿš€ Come experience it till May 3: ๐Ÿ“ Booth A410 (Hall A4) #PhysicalAI #AutoChina2026
1
2
873
Happy to share our latest from #AutoChina2026: DeepRoute Physical AI ๐Ÿš€ Our 3 Leaps: ๐Ÿง  Cognitive: 40B Foundation Model โšก๏ธ Efficiency: 10x Faster AI Loop โš™๏ธ Organizational: Human-AI Synergy ๐Ÿ“ See it at Booth A410 (Hall A4) #DeepRoute #PhysicalAI
1
294
Auto China 2026 is here! ๐Ÿš— Join DeepRoute.ai as we unveil the next chapter of #PhysicalAI and our mass-produced #AutonomousDriving solutions. Peace of mind, woven into every journey. ๐Ÿ“ Booth A410, Hall A4, CIECC ๐Ÿ“… Apr 24 - May 3, 2026 See you in Beijing! ๐Ÿ‘‡ #AI
1
3
214
Foundation Model: Redefining the "Brain" of Autonomous Driving. Here's our story from GTC 2026 in San Jose. ๐Ÿงต๐Ÿ‘‡ #GTC2026 #FoundationModel #AutonomousDriving #DeepRouteAI
1
145
The 40B Foundation Model โ€” a unified "brain" for autonomous driving. Perception. Reasoning. Action. Seamlessly integrated. And by cutting the data cycle from 5 days to just 12 hours, we've unlocked a new paradigm for scalable AI. โšก #FoundationModel #AI #DeepRouteAI
1
66
Proven at scale: ๐Ÿ“Š ~40% monthly market share ๐Ÿš— 250,000+ vehicles delivered ๐ŸŽฏ Targeting 1M+ by 2026 From Urban Autonomous Driving to Robotaxi โ€” the ultimate purpose of technology: peace of mind on every journey. #GTC2026 #PhysicalAI #AutonomousDriving
58
Insights from #NVIDIAGTC 2026! ๐Ÿš€ DeepRoute.ai is scaling the future: ๐Ÿ”นIn AI We Trust, In Scaling We Thrive. Redefining AD boundaries. ๐Ÿ”น10x efficiency: Data cycle compressed to 12 hours. Explore the next frontier of autonomous driving. ๐Ÿš— #DeepRouteAI
1
120
๐Ÿš— Our CTO Tongyi Cao speaks at #NVIDIAGTC 2026: Redefining Autonomous Driving with a 40B VLA Foundation Model. 200K+ vehicles. One flywheel: Model ร— Data ร— Simulation โ†’ L5. DeepRoute.ai = Future Mobility in Action ๐Ÿ“… Mar 17, 4 PM Beijing / 1 AM PDT ๐Ÿ“ Booth 1744
1
2
141
When AI begins to understand the world, driving becomes about anticipation, not reaction. Construction signs Text-based road guidance Human gestures Blind spots & irregular obstacles VLA (Vision-Language-Action) models turn perception into reasoning, and reasoning into safety.
176
Automotive KOL E-CoreChannel test-drove a production vehicle with DeepRoute.aiโ€™s VLA model in a Guangzhou urban villageโ€”one of the toughest urban environments. One take. No edits. VLA is already in mass production.#autonomousfuture #AI #LLMs
1
2
326
DeepRoute.ai at #CES2026 ๐Ÿš—๐Ÿค– โ€ข VLA-based autonomous driving demoed at Qualcomm booth โ€ข 3 GWM models equipped with DeepRoute.ai's urban NOA system โ€ข Recognised as NVIDIAโ€™s key L4 autonomous driving partner From innovation to scale #AutonomousDriving#ADAS
167