Andrej Karpathy

Tesla Autopilot Detects Cyclist Faster Than The Human Eye


Tesla Autopilot helped a driver detect a cyclist driving along the road faster than his human eye. Dirty Tesla shared his latest experience with Autopilot, revealing the company’s remarkable progress with its driver-assist software. 

Dirty Tesla was driving along a nearly empty road when the vehicle in front of him started swerving toward the middle of the lane. From the driver’s perspective, it was unclear why the vehicle ahead of him moving closer to the middle of the road since there didn’t seem to be any obstacles in the way. 

However, merely a few seconds later, Autopilot started ringing and warned Dirty Tesla about a cyclist along the side of the road. Autopilot’s alerts were a few seconds faster at detecting the cyclist than Dirty Tesla’s natural vision.

Drivers know that a few seconds is all it takes to go from a nice afternoon drive to an unexpected accident. In this case, Dirty Tesla’s extra pair of eyes helped keep the afternoon drive safe for everyone on the road that day. 

Tesla’s Senior Director of Artificial Intelligence, Andrej Karpathy, explained the company’s vision-based approach with Autopilot in a video from the Matroid Scaled Machine Learning Conference 2020. Other companies developing autonomous vehicles, like Waymo, mostly rely on Lidar. 

Elon Musk believes that Lidar will be a dead-end for companies trying to build a fleet of autonomous cars. Some may argue that Lidar is more reliable. However, ARK Invest defended Tesla’s approach towards autonomy. 

“LiDAR recognizes images and assesses direct depth more accurately than do cameras. In a two-step indirect process, cameras take shots of images and software gauges depth by analyzing the pixels. Mistakes involving a few pixels can translate into meters or yards of inaccuracy. By labeling 3D videos of driving scenes, Tesla is compensating for the camera’s weakness as the primary image sensor in its vehicles,” explained ARK Invest in its newsletter

Based on Tesla’s past explanations of Autopilot’s vision-based approach, the company has opted to take a more natural way of gathering data for its autonomous fleet. Dirty Tesla’s recent Autopilot experience showcased how far the company has progressed with its unique take on autonomous vehicles. It seems that Tesla has found a way to utilize the idea of vision and enhance it with Autopilot. 

Tesla’s progress with Autopilot now will undoubtedly improve its Full Self-Driving software in the long run. ARK Invest predicted that Tesla’s Robotaxi fleet could enjoy a near-monopoly in the autonomous ride-hailing service market because of the EV automaker’s unique approach towards autonomy.

The EV automaker’s vision-based approach, 3D labeling technique, and approach towards local mapping differentiated so much from its competitors, which could make Tesla the leader in the autonomous ride-hailing service market. 

Featured Image Credit: Dirty Tesla/YouTube

About the Author

Ma. Claribelle Deveza

Ma. Claribelle Deveza

Longtime writer and news/book editor. Writing about Tesla allows me to contribute something good to the world, while doing something I love.

Follow me on X

Reading next

Tesla Energy's Autobidder Will Be A Game Changer For Grid-Scale Battery Systems

Tesla Accessories