Avride, which spun off from Yandex, showed a four-wheeled robot courier on an Nvidia chip

The American company Avride, which separated from the Yandex driverless transport business, presented an updated version of its robot courier. It moves on four wheels instead of the previous six, and uses the Nvidia Jetson Orin artificial intelligence platform for navigation.

Image source: Avride

Engineers at Austin, Texas-based Avride said the previous-generation rover had many shortcomings: excessive tire wear caused by braking too hard on corners, insufficient maneuverability and insufficiently accurate execution of a given trajectory. In the new model, all relevant indicators have been improved.

The wheels of the updated delivery robot are mounted on movable arms attached to a rotating axis – when turning, each wheel moves around this axis, and the central lever stabilizes its position. “This design will allow the wheels to rotate both inward and outward, reducing friction when cornering,” explained Avride. By eliminating the traditional front and rear transverse axes, the robot is able to turn 180° almost instantly. This will come in handy when driving on narrow sidewalks, where you sometimes have to desperately maneuver; and when parking on slopes, where the wheels can be installed crosswise. Finally, tighter control helped increase the maximum speed of the machine, and therefore reduce delivery time.

The updated Avride robot courier not only received a new design, but also became smarter – Nvidia Jetson Orin platform, designed for powerful neural networks on unmanned vehicles, is now responsible for its navigation. It helps the machine process large amounts of data from lidars and cameras, with the help of which the robot navigates in space. Finally, the cargo compartment has become removable – it can now be modularly replaced for various purposes. The standard cargo area can accommodate several large boxes of pizza and drinks or several bags of groceries. A sliding lid has also been added, which ensures access only to the desired section, to avoid delivering products to the wrong customers.