Geographical Zone Vehicle Control for Occlusion-Aware Route Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Self-driving vehicles face challenges in navigating complex environments due to limited sensor fields of view, occlusions, sensor malfunctions, and reliance on user input, which can lead to safety risks and performance deviations.
Innovation Solution
A predictive vehicle control system using a neural network model to analyze traffic information and generate predictive video frames, predicting unsafe behaviors of moving objects and suggesting alternate routes to avoid collisions and other hazards, even when objects are outside the vehicle's field of view.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If self-driving vehicles rely on onboard sensors to learn and determine the surrounding environment, then the vehicle can take driving decisions in real time, but the limited Field of View of sensors may lead to occlusions and safety risks
Solution Approach 1:
The patent combines multiple sensor types (cameras, LiDAR, radar) to create a comprehensive sensing system that overcomes the limited Field of View of individual sensors. By merging data from multiple sources and perspectives, the system achieves both real-time processing and enhanced safety coverage.
Solution Approach 2:
The system implements a multi-functional sensing architecture where sensors serve multiple purposes: primary detection, redundancy backup, and predictive analysis. This universal approach allows the same sensor network to handle both immediate decision-making and long-term safety prediction.
2Adaptability or versatility
If sensors get dirty or malfunction, then the self-driving vehicle may rely on user input or other functioning sensors, but this may affect riding experience, risk safety, or result in performance loss
Solution Approach 1:
The patent implements predictive analytics that continuously monitor sensor health and environmental conditions to anticipate potential failures before they occur. By detecting early signs of sensor degradation or malfunction, the system can proactively switch to redundant sensors or alert the user, preventing safety incidents.
Solution Approach 2:
The system employs continuous feedback loops that monitor sensor performance in real-time, comparing expected versus actual readings. When anomalies are detected indicating dirty or malfunctioning sensors, the feedback mechanism triggers automatic compensation strategies using other sensors or alerts the user for intervention.
3Productivity
If the self-driving vehicle relies on user input when sensors malfunction, then the vehicle can continue operation, but this may potentially affect riding experience and risk safety
Solution Approach 1:
The system performs preliminary actions by maintaining multiple operational modes ready to switch between automatically. When sensor issues are detected, the vehicle has already prepared alternative operating procedures, allowing seamless transition without requiring user intervention, thus maintaining both productivity and ease of operation.
4Extent of automation
If self-driving vehicles use onboard sensors to determine the surrounding environment, then the vehicle can navigate autonomously, but occluding structures or intersections may limit sensor effectiveness
Solution Approach 1:
The patent utilizes multiple spatial dimensions and perspectives for environmental sensing. By combining data from sensors positioned at different locations and angles on the vehicle, and integrating information from multiple vehicles in the vicinity, the system reconstructs a complete environmental model that overcomes occlusions in any single viewpoint.
Data Source
AI summary
A control system and a method for vehicle control in geographical control zones is provided. The control system receives traffic information, including a plurality of image frames of a group of moving objects in a geographical control zone and generates a set of images frames of a first moving object of the group of moving objects based on application of a trained Neural Network (NN) model on the received traffic information. The generated set of image frames corresponds to a set of likely positions of the first moving object at a future time instant. The control system predicts the unsafe behavior of the first moving object based on the generated set of image frames and generates first control information, including an alternate route for a first vehicle in the geographical control zone based on the predicted unsafe behavior. The first vehicle is controlled based on the generated first control information.