As the auto industry prepares for a self-driving future, automakers must choose from a range of innovative technologies to develop safe and affordable autonomous vehicles (AV).

In the early phases of development, companies pursuing “full-stack,” level 4 and 5 autonomy favored a hardware-centric approach to solve the challenges of perception. Servers filled car trunks, requiring specialized cooling; unsightly LIDAR sensors costing hundreds of thousands of dollars were mounted to the tops of vehicles; and multi-chip radar arrays used digital processing on expensive FPGAs. While these high-end approaches helped demonstrate that complex perception challenges could be solved, the use of costly hardware made the implementation prohibitively expensive for commercial and consumer markets. The hardware sensors and associated computing often cost several times the cost of the vehicle.

Oculii’s Virtual Aperture Imaging (VAI) technology platform offers a software alternative to solve the hardware problem and make autonomous technology more accessible for commercial and consumer applications. Oculii uses artificial intelligence (AI) software to enhance commercial, market-proven, mass-manufactured radars, achieving the sensor perception required for autonomous operation at price points orders of magnitude cheaper than competing approaches. To demonstrate the capabilities of the VAI platform to improve performance by 100x, Oculii has released two full-stack radar products running on standard silicon radar platforms: EAGLE and FALCON.


EAGLE is the highest resolution commercial 4D imaging radar, with joint 0.5-degree horizontal and 1-degree vertical spatial resolution across a 120-degree horizontal and 30-degree vertical instantaneous field of view. Operating with radar sensors in the 76 to 81 GHz automotive band, EAGLE generates images with tens of thousands of pixels per frame and tracks targets to more than 350 m range, enabling vehicles to safely operate at high speeds and in all weather conditions. Its target sensitivity enables detecting and tracking low radar cross section targets like pedestrians and motorcyclists, making EAGLE well suited for use in congested urban environments. As required for automotive applications, the sensor operates across an extended temperature range from -40°C to 105°C.

Used on a low-power, two chip hardware platform with six transmit and eight receive channels, EAGLE provides the spatial resolution of an eight IC cascaded radar, i.e., with 24 transmit and 32 receive channels. It cost-effectively improves radar performance by more than 50x with just software (see Figure 1). Using DSPs for processing, which are lower cost and consume less power than FPGAs, EAGLE consumes less than 5 W biased at 12 V. It is smaller than an index card, enabling integration that preserve the aesthetics of the car (see Figure 2).

Figure 4

Figure 1 Comparing the Velodyne LIDAR (left) and Oculii EAGLE radar (right) imaging.

Figure 5

Figure 2 EAGLE radar attached to the front car bumper.


Smaller than a business card, Oculii’s FALCON is the most compact 4D imaging radar, with three transmit and four receive channels on a single IC. Also operating in the 76 to 81 GHz automotive radar band, it has a range of 200 m with 2-degree horizontal and 5-degree vertical resolution across a 120-degree horizontal and 30-degree vertical field of view.

Like EAGLE, FALCON provides long range coverage with high resolution across a wide field of view and operates from -40°C to 105°C. However, its power consumption is only 2.5 W. This, combined with small size, makes it well suited as a corner radar for advanced driver assistance systems (ADAS) or for autonomous robots, where low energy use is critical. Multiple FALCON sensors can be fused to cover all directions and work in all environmental conditions.


Until now, the only way to increase radar resolution has been to add more antennas, significantly increasing cost, size and power consumption. Oculii’s platform provides an alternative by using AI software, specifically an adaptive phase-modulated waveform that changes in real time with the environment. No additional antennas are needed. This dramatically improves radar resolution, increases range and widens the field of view without changing the bill of material or adding hardware cost.

Oculii’s AI software improves over time, learning from the environment as the radar sensors are exposed to more scenarios. The software adaptively embeds information from the environment, making the system “smarter” and improving resolution, range and sensitivity.

What makes Oculii’s software unique for AVs: it does not require new hardware. The sensors use the existing radar to improve sensor perception at a price point that makes autonomy more affordable. By solving the long-standing hardware limitations, Oculii can accelerate the arrival of AVs to the mass market.

Oculii’s products are also suited to the current automotive radar market. The pandemic has strained car production, adding financial and operational challenges to already costly and complex ADAS technologies. With radar sensors in millions of vehicles, Oculii’s AI-powered radar can use these same platforms to improve resolution over a wider field of view and extend detection range without changing hardware or requiring a structural redesign of the car.

Beavercreek, Ohio