In its quest to roll out the country’s first commercialized fleet of autonomous trucks, TuSimple has added an improved camera and vision system that lets its vehicles operate at night and in low-light situations. Currently, TuSimple can run its trucks for an average of 12 hours per day, or 50 percent usage, but this new system can push that up to an average utilization of 80 percent. It’s not difficult to do the math on how much more ground each vehicle can cover with what is essentially night vision on board.
Sensing technology is core to any driverless vehicle, but whereas other companies use lidar to map and measure obstacles by bouncing light off of them, TuSimple is focused on a more comprehensive set of technologies. Using an eight-camera array paired with both lidar and radar, the system sees in 360 degrees with a 1,000-meter detection range — a substantial advantage over the 250-300 meters afforded by systems that use lidar alone. TuSimple claims that its vehicles have three-centimeter control precision even in inclement weather.
The improved camera and vision system is proprietary and uses an automotive-grade CMOS image sensor from Sony Semiconductor Solutions (a Sony subsidiary) that relies on TuSimple’s software. The benefits extend beyond night driving, promising additional safety in the face of a wide variety of lighting scenarios. “Low light” conditions include dark tunnels, for example, and the system addresses the challenges presented by sun flare at either end of the day, headlights, traffic lights, and flickering from digital signage.
Built to meet the standards of Automotive Safety Integrity Level (ASIL)-C (ISO 26262), TuSimple’s camera array will be tested, developed, and manufactured by Sunny Optics.
“We weren’t able to find a camera system on the market that fit our needs, so we created one,” TuSimple president Dr. Xiaodi Hou said in a statement. Hou claims that TuSimple’s perception system “sees better than the human eye,” a statement he’ll no doubt expound upon at Nvidia’s GPU Technology Conference (GTC) this week in San Jose. Nvidia has invested in TuSimple, and not merely financially; the trucks rely on the chipmaker’s Drive AI hardware and software platform.
TuSimple’s self-driving trucks are class 8, level 4 fully autonomous vehicles, per the Society of Automotive Engineers’s (SAE) guidelines. The company is based in San Diego, where it has been building vehicles since 2015, but it operates its fleet from Tucson, Arizona.
And it has been on a fundraising tear of late, just a month ago adding another $95 million from Sina and Composite Capital to bring its total capital raised to $178.1 million. This leaves TuSimple with a $1 billion pre-money valuation. One of the stated purposes of the latest funding round was for projects with OEM, Tier 1, and sensor partners, and this camera project with Sony Semiconductor Solutions seems to fall under that purview.