Issue 37 Unmanned Systems Technology April/May 2021 Einride next-gen Pod l Battery technology l Dive Technologies AUV-Kit l UGVs insight l Vanguard EFI/ETC vee twins l Icarus Swarms l Transponders l Sonobot 5 l IDEX 2021 report

26 the Orin is designed to enable vehicle developers to scale their capabilities from SAE Level 2 up to Level 5. It can also be programmed through open CUDA and TensorRT application programming interfaces (APIs) and libraries, enabling the evolution of multiple product generations on the same platform. Hallgren notes that the Orin is well- known among Einride’s technology partners, making collaboration much easier than it would otherwise be. He also stresses the importance of the AI accelerators and the option of using CUDA operations for demanding algorithm calculations, which provides the flexibility needed to accommodate new functionality as customer demands evolve. The Orin can handle inputs from multiple high-resolution sensors including Lidars, cameras, GNSS receivers, IMUs and odometers, and Einride is working with a number of suppliers of these devices. Hallgren emphasises that this approach enables the company to strike the right balance between performance, integrity, robustness and maturity for its applications. The sensing suite continues to evolve, with the component sensors selected because they provide the perception needed for safe autonomous operation within the company’s specified ODDs. Hallgren says, “We also make sure the chosen sensor modalities complement each other in terms of fault modes and operating conditions.” Einride does most of the sensor integration in-house. “We typically write the sensor SDKs for reading and parsing sensor data, to ensure the best integration of that data in our autonomous systems stack,” Hallgren says. The company has also made several sensor SDKs available as open source software through its GitHub repositories (see sidebar). Written in the Go open source programming language, they include kits for Velodyne’s VLP-16 Lidar, Emlid Reach GNSS receivers and Xsens MTi IMUs, along with other systems such as Aplicom telematics devices, the Here April/May 2021 | Unmanned Systems Technology Dossier | Einride next-gen Pod Cooperative sensing shares redundant measurements (10, 12) of an object’s position (201) made by connected vehicles (101, 102). Each has sensing (2), localisation (3), communication (4) and control (5) systems. The object’s position is translated into each vehicle’s local coordinate system (15, 16), with errors reduced by a probabilistic filter (52) in each vehicle’s perception system (50) We typically write the sensor SDKs for reading and parsing data, to ensure the best integration of that data in our autonomous stack