Lotus uses Ambarella AI radar in L2+ systems for Eletre and Emeya EVs

The vehicles’ L2+ semi-autonomous systems that have benefited from this radar technology include highway and urban navigation on autopilot (NOA) and automatic emergency braking (AEB), where the ultra-long detection range of over 300 meters provides more time to safely react to vehicles and other objects while traveling at highway and racetrack speeds. These systems were developed by Lotus Robotics, a wholly owned subsidiary of Lotus, as part of its autonomous driving platform.

Ambarella’s Oculii radar technology is providing ultra-fine angular resolution of one degree in both vehicles, using only six transmit and eight receive antennas on each of the vehicles’ two radar modules – more than double the resolution of the nearest 4D imaging radar competitor using the same number of antennas. This high angular resolution is important for clearly distinguishing, identifying, and locating objects and people, both for nearby detection in crowded urban environments, as well as from a far distance when travelling at highway and racetrack speeds, to ensure timely and accurate avoidance and braking by the vehicles’ L2+ systems.

This high resolution, in combination with a low antenna count, also provides higher value, while significantly reducing radar system power consumption and complexity; enabling the radar modules to be fitted into the desired locations on the vehicles. This combination can only be achieved using Ambarella’s Oculii AI radar algorithms, because of their unique ability to adapt radar waveforms to the environment. In fact, other 4D imaging radars require double the number of antennas to achieve the same level of performance, and in some cases, even more; leading to at least double the power consumption.

“The performance that Lotus has achieved using our Oculii 4D imaging radar technology provides evidence for the real-world benefits of our AI innovations that adapt radar waves to the surrounding environment,” said Fermi Wang, President and CEO of Ambarella. “Building on their current models’ significant achievements with edge-processed AI 4D imaging radar, Lotus is evaluating Ambarella’s centralised radar processing architecture to realise even greater ADAS and autonomous performance.”

By centrally processing raw 4D imaging radar data and fusing it at a deep level with the vehicle’s other sensor information, Ambarella’s AI software-defined architecture is capable of detecting objects over 500 meters away, while providing the ability to shift processing power among sensors and adapt to real-time driving conditions. This architecture also enables even higher angular resolution of 0.5 degrees, and an ultra-dense point cloud with tens of thousands of detection points per frame. Additionally, because Ambarella’s Oculii AI radar algorithms uniquely adapt radar waveforms to the environment, an order-of-magnitude fewer antennas is still utilised for centralised processing, maintaining reduced data bandwidth and power consumption compared to competing 4D imaging radar solutions.

Test drives demonstrating Ambarella’s centralised radar processing architecture will be available during AutoSens Europe, taking place 8–10 October in Barcelona.