Automated Vehicle Handling Unknown Objects via Cloud Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Automated vehicles face challenges in handling unknown objects in their environment, as existing technologies struggle to recognize and respond to rare or unfamiliar objects, leading to potential vehicle stops and the need for human intervention.
Innovation Solution
A method and system that utilize a sensor system to detect unknown objects, search databases for similar objects based on environmental data, determine typical properties, and provide recommendations for action to the vehicle, allowing it to safely navigate without human intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If algorithmic or artificial intelligence-based methods are used to detect objects, then the vehicle can automatically identify common objects, but unknown or rare objects cannot be detected
Solution Approach 1:
The system performs preliminary actions by capturing images of unknown objects with image acquisition units before making driving decisions. These images are then analyzed by external servers using machine learning models to identify the objects and determine appropriate vehicle responses, allowing the system to prepare for potential hazards in advance rather than reacting too late
Solution Approach 2:
An external server acts as an intermediary between the vehicle's image acquisition units and the control unit. The server receives images of unknown objects, processes them using trained machine learning models, and returns identification results and recommended actions to the vehicle, enabling the system to handle object types that were not present in the original training data
2Reliability
If the vehicle stops to wait for human driver instructions when unknown objects are detected, then safety is maintained, but productivity and efficiency are reduced
Solution Approach 1:
The system implements a feedback loop where the control unit receives continuous information from image acquisition units about unknown objects, processes this information through the external server, and automatically adjusts driving decisions based on the returned recommendations. This closed-loop feedback system enables automated vehicles to independently handle unknown objects without requiring human intervention, thereby maintaining both safety and productivity
Solution Approach 2:
The automated vehicle performs self-service by autonomously detecting, analyzing, and responding to unknown objects using its own image acquisition units, external server analysis, and control unit. The system serves itself by making independent driving decisions based on real-time object identification, eliminating the need to stop and wait for human driver instructions
3Adaptability or versatility
If comprehensive object detection is attempted to cover all possible objects, then detection coverage improves, but device complexity and computational requirements increase
Solution Approach 1:
The system achieves universal object detection capability by using a trained machine learning model on an external server that can identify multiple types of objects across different categories. This single multi-functional system replaces the need for multiple specialized detection algorithms, thereby maintaining high detection coverage while reducing overall system complexity
Solution Approach 2:
The system transitions from local on-vehicle processing to cloud-based analysis, moving the computational dimension from the vehicle to an external server. This dimensional shift allows comprehensive object detection to be performed with powerful remote computing resources while keeping the vehicle's onboard system relatively simple, effectively separating detection complexity from deployment constraints
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to a method for supporting an automated driving vehicle, wherein objects in the vehicle's environment are detected by means of sensors (20), and wherein objects are identified and recognized in the detected environment data (21) by means of an object recognition device (22), wherein, if an unknown object (12) is present in the environment, the following steps are performed: (a) searching for the unknown object (12) in at least one database (4) by means of a search device (2), (b) determining typical properties (10) of the unknown object (12) based on the search result (6) by means of the search device (2), (c) deriving a recommendation for action (24) for the automated driving vehicle based on the typical properties (10) of the unknown object (12) by means of a recommendation device (3), (d) providing the derived recommendation for action (24).Furthermore, the invention relates to an associated system (1).