Robotaxi sensor platform promises „safety comparable to aviation“
Founded in 2017 and at home in California, NPS is pursuing a vision of traffic with “zero accidents.” According to CEO Behrooz Rezvani, this will actually only be possible with autonomous cars – cars in which the driver no longer has to continuously monitor the systems and prove his attendance by regularly touching the steering wheel or the like. In other words, with vehicles that reach at least level 4 of the SAE definition of autonomous driving, in which the driver can quite easily take a nap or read the newspaper. Or, of course, with vehicles of the highest autonomy level 5, which manage entirely without a steering wheel.
However, such vehicles require very sophisticated technology. A technology that is very close to the state of perfection. For today, accidents involving autonomous cars generate much more negative media attention than those involving human drivers – think of the fatal accidents involving autonomous driving Tesla and Uber vehicles in recent years. “Society forgives human errors in traffic – but not technology errors,” Rezvani said.
That’s why NPS is stepping up to the plate with the ambition of ensuring driving safety at least on par with the safety level we see in commercial aviation. In its first product, the NPS 500 sensor platform, the NPS development team led by CTO Babak Hassibi has realised corresponding conceptual assumptions: Camera, radar and lidar work together; their effects partially overlap, increasing the reliability of the data and thus the necessary redundancy. Insofar, not a big surprise. However, a second pillar of the concept is based on the consideration that the automatic systems detect a potentially dangerous situation earlier. This gives the system more time to react. This increased time buffer is achieved through a greater range of the sensors. “We consider a sensor range of 500 metres to be essential,” explains Rezvani.
But that’s not all: the NPS 500 can also “see around the corner” with its sensors and it can quickly and correctly detect very complex situations with many road users and chaotic movement patterns.
For this purpose, NPS uses, for example, a multi-band radar which, in contrast to radar sensors commonly used today, transmits simultaneously on several frequency bands – one frequency range of the radar is below 1 GHz, a second at 5-6 GHz and a third in the low millimetre wave range at 24 GHz. The signals from this multi-band radar are processed together using triangulation methods, and specially developed AI algorithms generate the necessary accuracy and resolution, promises Rezvani.
The company is also going its own way with the lidar sensor. The platform is equipped with a solid-state lidar that transmits and receives multiple infrared beams with an extremely high angular resolution of 0.05° via MIMO mechanisms. The whole thing takes place at a high frame rate. This is supplemented by several cameras. Together with a signal processing system developed in-house, Rezvani wants to achieve a much higher performance and reliability of the generated data – the NPS founder speaks of 70 times higher resilience.
Based on this technology, NPS has developed a system with 360° all-round vision that is mounted on the roof of a vehicle, which of course limits the design freedom of the vehicle designer. However, the NPS 500 is likely to play in a league where relevant vehicle studies and prototypes today are also equipped with a rotating lidar sensor, typically mounted on the roof. Compared to the rotating lidar sensors that are commonly used here, however, Rezvani expects a much lower price tag – talk of around $5000.
Which submarket is Rezvani targeting? Autonomous vehicles in Levels 4 and 5, of course, but the major European and American carmakers are not there yet – they are currently mostly working on Level 2+ or L3 capabilities. “Level 3 will not take off,” Rezvani is convinced. The reason: Level 3 requires the driver to take command within seconds should the system fail or a driving situation arise that overtaxes the electronics. This requirement is not really a benefit to the driver. “Most OEMs will therefore skip Level 3 and move straight to L4,” says Rezvani.
And who is behind the company? NPS is funded by venture capitalists Cota Capital and Blue Lagoon. There are also a number of smaller investors, but they “prefer not to be named” at the moment.
More information: https://nps.ai/
4D automotive imaging radar slashes costs by two thirds
Bosch, Continental invest in AI vision startup
AEye, Continental jointly develop long-range lidar
AutoX sends robotaxis into public service
Mobileye gives a glimpse of its future technology
Compact lidar sensors enable 360° all-round vision