Inside Tensor's Truly Self-Driving Robocar (w/ Tensor COO Jewel Li)
Get the full intelligence
Search transcripts, export clips, track mentions, and explore all topics from “Inside Tensor's Truly Self-Driving Robocar (w/ Tensor COO Jewel Li)” inside PodZeus.
In this episode of Ride AI, host Sophia Tung interviews Jewel Li, COO of Tensor, about the company's groundbreaking Tensor Robocar—the first consumer-ready Level 4 self-driving vehicle designed from the ground up. Li shares her journey from engineering at Wuhan University and a PhD at the University of Delaware, to working on IBM Watson’s DeepQA system that defeated humans on Jeopardy. She discusses the evolution of autonomous driving, emphasizing the shift from retrofitted vehicles to purpose-built, sensor-rich designs. The Tensor Robocar features a unique 37-camera, 5-LiDAR, 11-radar, and 22-microphone sensor suite, including under-chassis cameras to solve the 'cold start' problem. The car’s 8,000 TOPS compute system, with dual redundant AI layers, ensures safety and future-proofing. A retractable steering wheel balances regulatory concerns with user experience, while the design prioritizes passenger comfort with reclining seats, a Dolby Atmos sound system, and a long, spacious cabin. Li reflects on Tensor’s origins as AutoX, its pivot from grocery delivery and robotaxis to a consumer-focused robocar, and the strategic decision to launch first in the UAE before expanding to the U.S. in 2027–2028. The company’s confidence in its timeline stems from five years of development and rigorous validation. Key takeaways include: 1) Purpose-built, ground-up design is essential for scalable, safe Level 4 autonomy; 2) Sensor redundancy and multi-modal perception (cameras, LiDAR, radar) are non-negotiable for safety; 3) The retractable steering wheel is a UX and safety innovation that addresses the risks of human override in full autonomy; 4) Massive on-board compute (8,000 TOPS) is necessary to run powerful main and backup AI systems; 5) The cold start problem—ensuring safety when the car is first powered on—requires under-chassis sensors; 6) Regulatory readiness in markets like the UAE, with Vision 2030 ambitions, is a key enabler; 7) The business model includes both personal ownership and fleet sharing via partnerships like Lyft and Green Mobility in Denmark; 8) Consumer demand for a truly self-driving car is growing, especially as the technology proves reliable and safe.
Ground-up design is essential for scalable, safe Level 4 autonomy, unlike retrofitted vehicles.
Sensor redundancy across multiple modalities (cameras, LiDAR, radar) is critical for safety and reliability.
The retractable steering wheel balances regulatory requirements with user experience and safety.
8,000 TOPS of on-board compute ensures powerful main and backup AI systems, future-proofing the vehicle.
Under-chassis sensors solve the 'cold start' problem—detecting hazards when the car is first powered on.
…and 3 more takeaways available in PodZeus
Introducing the Tensor Robocar and Jewel Li
Host Sophia Tung introduces Jewel Li, COO of Tensor, and sets the stage for a deep dive into the company's first consumer-ready Level 4 self-driving vehicle, the Tensor Robocar. Li’s background in AI and autonomous systems is highlighted, including her work on IBM Watson’s Jeopardy-winning DeepQA system.
From IBM Watson to Autonomous Driving
Li recounts her early fascination with AI, particularly natural language processing and tensor operations, which led her to pursue a PhD. She reflects on the historic moment when Watson beat human champions on Jeopardy, a milestone that drew her to IBM’s core AI teams.
The Evolution of Autonomous Driving at AutoX
Li discusses her nearly decade-long journey building self-driving companies since 2017, starting with AutoX in a house in Saratoga using Logitech cameras. She reflects on the industry’s shift from early prototypes to volume production and the challenges of retrofitted vehicles.
Sensor Architecture: The 37-Camera, 5-LiDAR System
“We have this cold start problem when the car is first turned on, we need to check, the system needs to check, not us, not our employees, but the system needs to be able to check what is underneath the chassis.”
Why Not Just Cameras? The Case for Multi-Modal Sensing
“There's nothing, there's no price on safety.”
“There's nothing, there's no price on safety.”
“In level four, you are allowing people not to look at the road, right? You're allowing people to get to sleep. Then you can't have that level two easy disengagement because people can... disengage by not even knowing that they disengaged, by accident.”
“We have this cold start problem when the car is first turned on, we need to check, the system needs to check, not us, not our employees, but the system needs to be able to check what is underneath the chassis.”
Host
Guest
Jewel Li
person
Tensor
organization
AutoX
organization
UAE
place
IBM Watson
organization
Lyft
organization
Jeopardy
media
Green Mobility
organization
Waymo
organization
University of Delaware
organization
Get the full intelligence
Search transcripts, export clips, track mentions, and explore all topics from “Inside Tensor's Truly Self-Driving Robocar (w/ Tensor COO Jewel Li)” inside PodZeus.
Start discovering podcast insights today
Start with a 7-day trial and explore a growing catalog of popular podcasts. No credit card required.
No credit card required • 7-day trial • Cancel anytime
