Rear-Facing Camera Pedestrian Detection and Motion Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles and driver-assistance systems face challenges in detecting and predicting pedestrians early enough to avoid collisions, particularly in environments like intersections where pedestrians may disobey traffic signals or misjudge situations, leading to potential collisions before they can be perceived by sensors or drivers.
Innovation Solution
A safety system incorporating a rear-facing camera with a wide viewing angle and a pedestrian component that uses a two-stage computer vision-based deep learning technique to detect and predict pedestrian motion, employing saliency maps and neural networks to identify potential pedestrian locations and track their behavior, even when they are out of the camera's view, allowing for timely warnings or evasive actions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If forward-looking sensors are used for pedestrian detection, then the system can detect pedestrians in the vehicle's path, but pedestrians may enter blind spots or cross paths before being detected
Solution Approach 1:
The system performs preliminary pedestrian detection by using rear-facing cameras to detect pedestrians before they enter the vehicle's blind spots or cross paths. The pedestrian component analyzes images from rear-facing cameras to identify pedestrians in early stages, allowing the system to take preventive action before a collision risk arises. This preliminary detection approach addresses the timing issue by detecting pedestrians earlier in their movement trajectory.
2Loss of time
If rear-facing cameras are used to detect pedestrians early, then detection time is improved, but the system complexity increases due to additional sensors and processing
Solution Approach 1:
The pedestrian component is designed to perform multiple functions using the rear-facing camera system. It not only detects pedestrians but also determines their motion, predicts their future positions, and integrates this information with data from other sensors. This multi-functional approach reduces the need for separate dedicated systems, thereby managing complexity while achieving early detection.
Solution Approach 2:
The pedestrian component acts as an intermediary that processes and integrates information from rear-facing cameras with other vehicle sensors. It serves as a mediator that combines data from multiple sources, performs coordinated analysis, and generates unified predictions about pedestrian behavior. This intermediary approach allows the system to leverage existing sensors more effectively without requiring completely separate detection systems.
3Measurement precision
If the system waits for pedestrians to be clearly visible, then detection accuracy is improved, but collision risk increases as pedestrians may be in blind spots
Solution Approach 1:
The system performs preliminary detection of pedestrians using rear-facing cameras before they enter blind spots or become difficult to detect. By identifying pedestrians early when they are still visible and can be accurately measured, the system maintains detection accuracy while preventing the situation where pedestrians are in blind spots. The pedestrian component analyzes visual features and motion patterns to confirm pedestrian identity and intent early in the detection process.
Solution Approach 2:
The system adds the temporal dimension to pedestrian detection by analyzing pedestrian motion over time and predicting future positions. Instead of relying solely on static detection accuracy, the system uses motion tracking and prediction to assess collision risk dynamically. This allows the system to make accurate assessments even when pedestrians are in transitional states, by considering their trajectory and predicted future positions relative to the vehicle's path.
Data Source
AI summary
Systems, methods, and devices for pedestrian detection are disclosed herein. A method includes receiving one or more images from a rear-facing camera on a vehicle. The method further includes determining that a pedestrian is present in the one or more images, predicting future motion of the pedestrian, and notifying a driver-assistance or automated driving system when a conflict exists between forward motion of the vehicle and the predicted future motion of the pedestrian.