According to Serve Robotics, an Uber spinout that develops sidewalk delivery robots, the company is deploying its next generation of robots capable of making certain commercial deliveries without the involvement of a human, according to the firm. Serve will no longer depend on remote operators to teleassist robots or followers to trail after robots for safety in specified operational design domains or geofenced zones.
The majority of companies in the industry, including Coco, Starship Technologies, and Kiwibot, rely on remote operators to monitor autonomous deliveries and take over driving if the robot breaks down or requires assistance, so Serve’s achievement is a significant step forward in the advancement of robotic deliveries.
In an interview with TechCrunch, Ali Kashani, co-founder, and CEO of Serve explained how the company solved the problem of relying on teleoperation for safety. “Relying on teleoperation for safety means you must rely on 100 percent reliable LTE networks and 100 percent mistake-free operators, both of which are impossible to achieve consistently,” Kashani said. Imagine a safety emergency requiring human intervention, but the video is delayed, or the link has been lost. What happens then? Humans are no longer need to be in the loop while using Level 4 robots to assure safety.”
Serve began rolling out its next generation of robots in December, and the company recently announced that it had completed its first delivery of Level 4 autonomy, which the Society of Automotive Engineers defines as a system that can drive autonomously as long as certain conditions are met and will not require a human to take over the steering wheel. In specific communities in Los Angeles, such as Hollywood, where Serve has been operating since 2018, Kashani said the startup’s robots presently have L4 capabilities. The company is looking to expand.
As Kashani explained to TechCrunch, “When the robot is in a certain location where Level 4 is enabled, the remote video feed is turned off, and the robot continues navigating autonomously without the need for a person in the loop.” “If the robots find themselves in a situation where they need help, such as when they come upon something unexpected, they may always seek assistance.” They also have their video cameras turned on when crossing junctions. “However, for the vast majority of the time, they work independently.”
There will inevitably be a long tail of edge situations that robots are unfamiliar with until autonomous cars achieve Level 5 competence. At that point, the system can function in all settings without a human. According to Kashani, relying on people for specific tasks makes sense from safety and profitability.
Several active sensors, including ultrasonic and lidar sensors from Ouster and passive sensors, such as webcams, are included in Serve’s latest robots, which can negotiate congested sidewalks with more ease. According to the company, to provide specialized capabilities for its bots, the business has built skills such as automated accident prevention, vehicle collision avoidance, and fail-safe emergency braking. The calculations required to achieve such capabilities in real-time are driven by the Jetson platform from chipmaker Nvidia, which is mainly developed for robots and other autonomous vehicles.
According to the firm, the $13 million enlarged seed round it received last month would help support its rapid development ambitions into new consumer groups and geographic locations. By those stated objectives, Kashani claimed that Serve’s next steps would be to deploy its next-generation robots in different places, beginning with a more significant presence in Los Angeles.