Perceptive changes everything. Not only can autonomous vehicles finally sense and make sense of the world as a human would, they can also be built at scale and be broadly deployed.
A single, integrated, centralized sensing unit housed inside the vehicle’s cabin, communicating with small, nearly passive, rugged, and inexpensive multimodal probes placed on the vehicle’s exterior. No more bulky protrusions spoiling vehicle designs, or vulnerable and expensive electronics exposed to the elements. And upgrading sensing capabilities through the latest algorithms is only one software update away.
A truly digital architecture that senses the environment by sending and receiving digital signals. Our platform allows autonomy developers to treat sensing as a computational problem solvable in the software domain, with regular GPUs and CPUs, and with complete control and ownership over algorithms. Perceptive’s software-defined approach streamlines the sensing hardware, dramatically simplifies vehicle integration, and enables novel and more robust solutions to be prototyped and deployed rapidly.
With our unique Concurrent SensingTM technology, the Perceptive platform retrieves multimodal signals from the environment (laser, RF, or optical) and blends them together into a single, ultra-raw sensory stream. Based on this type of input, concurrent ML algorithms produce a single, rich, and incredibly insightful point cloud of the environment that enables unmatched detection and perception capabilities.
Sensing is solved computationally, high-up in the software and compute stack – a fundamental departure from the conventional hardware-centric sensing approach.
Multimodal signals and sensing hardware are abstracted by an operating system. The sensing hardware is software-agnostic and the software is decoupled from specific hardware configurations.
Digital Sense is the unique technology enabling the Perceptive architecture. A centralized sensing hardware front end interrogates the environment with digital signals, channeled in and out by optical fibers attached to multimodal probes. The probes send and receive digital signals into the environment by encoding data packets in laser beams and microwaves – as is done in established wireless or optical fiber telecom systems.
Designed-top down with the ultimate goal in mind: human-like perception, which replaces the myriad of third-party devices in today’s autonomous vehicles with an integrated and harmonious solution.
Augments sensing performance, eases hardware requirements, and crucially allows the entire system to learn from experience – like humans.
Applications implemented high-up on the software stack. Sensing becomes a software engineering effort, open to endless, rapidly deployable iterations. Because engineering ingenuity – and not architecture – should be the only limit to the system’s capabilities.
Sensing methods inspired by powerful and scalable digital telecommunications systems well-established in wireless, optical fiber, and datacenter networks.
Accessible to user-defined software applications and built-in GPUs for machine learning training or inference. That’s all information that today’s autonomy sensors throw away.
Software that users may fully own and control. No software black boxes when human lives are at stake.