Three-Camera Electronic Mirror for Cyclist Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional rear view mirrors and electronic camera systems fail to provide early detection of on-coming cyclists in bike lanes, especially on curved roads, due to limited field of view and resolution issues with radar systems, leading to potential accidents.
Innovation Solution
A three-camera electronic mirror system that uses computer vision to detect moving objects, predict their path, and generate notifications before they are in the driver's field of view, utilizing a rear-facing camera to augment side camera views and prevent accidents by alerting the driver of approaching cyclists.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If a traditional reflective side view mirror is used, then the driver can see the side view, but on curved roads cyclists in bike lanes cannot be seen until very close to the vehicle
Solution Approach 1:
The patent uses a rear-facing camera positioned at the rear of the vehicle to capture images of cyclists approaching from behind on curved roads. This rear view dimension is then processed through computer vision to predict cyclist trajectories and provide early warning before cyclists enter the side mirror's field of view. This adds a temporal and spatial dimension to the traditional side view mirror system, allowing detection of cyclists before they become visible in the side mirror.
2Loss of time
If a camera is mounted to provide a similar view as the traditional side view mirror, then electronic mirror functionality is achieved, but advanced warning of approaching cyclists is not provided
Solution Approach 1:
The system performs preliminary detection of cyclists using a rear-facing camera before they enter the side mirror's field of view. Computer vision algorithms analyze the rear camera images to detect cyclists and predict their future positions. When the predicted path indicates the cyclist will approach the driver's side, the system generates an early warning notification, providing advance notice before the cyclist becomes visible in the side mirror view.
Solution Approach 2:
The patent introduces a rear view dimension captured by a rear-facing camera, which is then processed through computer vision to predict cyclist trajectories. This temporal-spatial dimension allows the system to warn drivers of approaching cyclists before they enter the side mirror's field of view, providing advanced warning that a traditional side-view camera alone could not achieve.
3Measurement precision
If blind spot detecting radar is used, then object detection capability is provided, but the radar lacks resolution to accurately detect and classify cyclists
Solution Approach 1:
The patent replaces radar-based detection with a camera-based visual detection system. The rear-facing camera captures images that are then processed using computer vision algorithms to detect, classify, and track cyclists. This optical approach provides superior resolution and classification accuracy compared to radar, while maintaining reasonable system complexity through the use of standard camera components and software-based image processing.
4Loss of information
If the vehicle body blocks the radar signal range on curved roads, then detection of cyclists behind the vehicle is enabled, but the radar cannot detect cyclists approaching from behind on curves
Solution Approach 1:
Instead of using side-mounted sensors that have their view blocked by the vehicle body on curved roads, the patent inverts the sensing approach by placing the camera at the rear of the vehicle. This rear-facing camera has an unobstructed view of cyclists approaching from behind, even on curved roads. The camera captures images that are then processed to detect cyclists and predict their trajectories, providing detection capability that overcomes the vehicle body blockage problem.
Data Source
AI summary
An apparatus including cameras and a processor. The cameras may generate pixel data of an exterior view from a vehicle. The processor may generate video frames from each of the cameras, perform computer vision operations on the video frames to detect an object, determine a predicted path of the object with respect to the vehicle, predict an approach side of the vehicle of the object based on the predicted path, and generate a notification in response to the predicted path. A first of the cameras may be on the approach side of the vehicle. The object may not be in a field of view captured by the first camera. The object may be detected in the video frames captured by a second of the cameras that may not be on the approach side of the vehicle. The notification may be generated before the object may be in the field of view.