Self-driving perception

This invention

This invention is an autonomous vehicle perception system that uses neural networks to detect and classify objects around a vehicle. It fuses data from cameras, LiDAR, and radar, then processes that combined information in real time to support navigation and driving decisions. It sits where machine learning, computer vision, and autonomous vehicle control meet — an area focused on giving a vehicle one reliable, unified picture of its surroundings.

Where it fits

Your idea lives in a well-developed corner of vehicle technology, touching several closely related areas: Autonomous Control (G05D), Radar & Navigation (G01S), Computer Vision (G06V), Vehicle Control Systems (B60W), and AI & Machine Learning (G06N). The results cluster strongly here. Autonomous Control (G05D) appears at roughly 62× the corpus baseline, and Radar & Navigation (G01S) at about 50×. That points to a focused, actively pursued direction. Filings climbed from 2019 onward, with nine in 2019 and eight in 2020, reflecting sustained interest. Active groups include ZOOX INC and MAGNA ELECTRONICS INC, each with multiple patents here, alongside GOOGLE INC, TOYOTA, and DELPHI TECHNOLOGIES INC. The field also connects to Traffic Control (G08G), which these results touch a little less.

Closest related work

US-11852746-B2 — Multi-sensor fusion platform for bootstrapping the training of a beam steering radar (Metawave Corporation · 3 citations · 2-member family · filed 2023, recent)

This patent describes a fusion platform with separate camera, LiDAR, and radar perception engines, each running its own neural network to detect and identify objects. It shows how Metawave Corporation approached the same multi-modal challenge you're tackling. The system keeps per-sensor neural networks but combines their results, with an interesting twist: it uses fusion to help train the radar. It's a close conceptual neighbor to your detect-and-classify-across-sensors design. Metawave has 2 patents in this result set.

US-11494937-B2 — Multi-task multi-sensor fusion for three-dimensional object detection (UATC LLC · 4 citations · 5-member family)

This patent covers machine-learned models that fuse multiple sensors for 3D object detection in autonomous perception and control. It emphasizes training a model ensemble across related tasks at the same time. UATC LLC frames the fusion problem as a learning architecture rather than a post-detection merge — a useful contrast to approaches that detect separately and combine afterward. UATC has 2 patents in this set.

US-11062454-B1 — Multi-modal sensor data association architecture (Zoox Inc · 56 citations · 3-member family)

This widely cited patent trains a machine-learning architecture to associate point-cloud data from LiDAR and radar with objects detected in camera images, and to generate 3D regions of interest. It shows how Zoox handled the core data-association question: matching what different sensors see to the same real-world object. That question is central to any fusion-based perception system like yours.

US-10599150-B2 — Autonomous vehicle: object-level fusion (The Charles Stark Draper Laboratory Inc · 17 citations · 2-member family)

This patent normalizes vision, radar, and LiDAR outputs into a common format, then fuses them at the object level, associating detections and predicting trajectories. It shows a contrasting object-level fusion philosophy — combining already-detected objects rather than raw signals or features. That gives you a clear view of one end of the design spectrum for multi-sensor perception.

What you can do next

• Explore & build on it. Browse the related work above — new, differentiated ideas often come from combining or improving on existing approaches (a specific fusion mechanism, a novel network architecture, a particular sensor-association method, or a real-time processing technique others haven't pinned down).
• If you'd like to protect it. Filing a provisional application (usually with a patent attorney) is a common first step. Most inventions can be protected in some form — what matters is how broad and defensible that protection is, which is where a patent attorney adds value (a very narrow claim may be granted but protect very little).
• If you'd like to make or sell it. The patents above point to who holds rights in this space; if your product would use a protected approach, licensing is a path worth exploring.