Introducing the Next-Generation Autonomous Sensing System

Mina Rezk

Realizing safe and scalable autonomous vehicles requires them to have superhuman-like awareness of their surroundings, making camera, LiDAR and radar technologies a basic necessity.

As such, LiDAR sensors today provide dense point clouds representing distance and reflectivity of objects and radar provides sparse measurements representing distance and velocity. Data from all sources is then fused in high power compute running machine learning and computer vision algorithms to identify objects on the road and safely maneuver the autonomous vehicle.

At Aeva, we are introducing a new type of sensing system that is fundamentally different to what is currently available. Our new proprietary 4D LiDAR technology possesses the best of both worlds, in one compact box. We are able to not only improve the performance of traditional LiDARs, but also give each LiDAR point an instantaneous measurement of velocity, unlocking the temporal element that is often overlooked and missed.

Aeva sensing system mounted on a vehicle streaming raw outputs

Crucially however, by simultaneously measuring dense position and dense velocity, we give our AI algorithms a whole new dimension of features, that have historically been locked away, and inaccessible. We combine the advantages of LiDAR (measuring depth and object reflectivity at long distance), radar (measuring instant velocity at an order of magnitude better spatial resolution and accuracy than conventional radar), machine vision, and highly-accurate motion sensors, into this single box.

Over the last one and a half year, we have implemented a rigorous product development process, informed by the collective experience that my co-founder, Soroush Salehian, and I have in automotive and consumer electronics, to build a technology that represents a generational leap for autonomous vehicles. Our sensor was developed from the ground up using proprietary hardware and software to be highly scalable and manufactured using mature chip-level technology.

Outputs of Aeva sensing system. Top left is the depth map where the color of each pixel represents the depth in the scene. Top right is the velocity map where the color of each pixel represents instant velocity. Red represents objects moving away from the sensor and blue represents those moving toward the sensor. Bottom left is the reflectivity map where the color represents the inherent reflectivity of the object material, where yellow represents high reflectivity and blue represents the low reflectivity. Bottom right is the vision map from the natively integrated camera sensor.

As outlined earlier, the heart of our system is our proprietary 4D LiDAR technology, which uses the spectrum of invisible light to measure the environment but operates similar in principle to the way Radar uses radio frequency to measure its surroundings. The spectrum of light is continuously embedded with a unique signature in multiple dimensions and then sent out to the environment, mapping the scene as the light bounces off of objects. A portion of this light is then captured with our optics, and its characteristics are compared with the outgoing beam to measure the range, velocity, and reflectivity of those objects. This technology enables a number of key differentiators unique to Aeva:

  • New dimensions for safer AI: Using our 4D LiDAR technology, we are able to simultaneously measure instantaneous velocity, depth and reflectivity for every point measured in the scene unlocking a whole new dimension for safe AI algorithms.
  • Immune to interference: A crucial but often overlooked aspect of traditional LiDAR systems today, is that their physics guarantees that they will interfere with each other. Our system is immune to interference from other sensors — and even from the sunlight itself — by design.
  • Long range performance at low power: Achieving single photon detection without the use of any exotic detectors. This allows us to measure low reflective targets beyond a 200m range at low power.
  • Unparalleled points per second: With our technology we can perform multiple measurements in the air without compromising range performance, which puts Aeva in the unique position to achieve a point density that approaches camera-like resolutions.
  • Scalable architecture: Our architecture is free of exotic components, which allows us to put the first 4D LiDAR on a single chip.

All of these differentiators provide a sensing system of unprecedented performance, giving vehicles equipped with our technology the complete insights they need to see, understand, and navigate their environment in a drastically safer way.

Aeva sensing system long range performance in a highway scene. On the left is a top view of the depth map; on the right is a top view of the velocity map. The top of the picture shows vehicles at 200m away from the sensor.

Autonomous technology will revolutionize the way the world drives, and we along with our key partners believe that our technology will play a big part in enabling autonomous vehicles to go mainstream. It’s just the beginning!

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade