Fleet Algorithm Selection for Reliable Autonomous Driving
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing autonomous driving systems in vehicle fleets face reliability issues due to outdated algorithms and inconsistent performance across vehicles with identical sensor technology, leading to frequent abortion of autonomous modes and insufficient warning times for drivers to take control.
Innovation Solution
A control device, external to the vehicle, compares algorithms of multiple vehicles with identical sensor technology to determine a most reliable algorithm, generates early warnings, and updates vehicles with the best algorithm to ensure consistent and reliable autonomous driving across the fleet.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a single motor vehicle uses an autonomous driving algorithm, then the vehicle can perform autonomous driving mode, but the reliability and accuracy of the driving mode vary severely depending on whether the system is up to date
Solution Approach 1:
The system collects feedback from multiple vehicles about their autonomous driving performance and system update statuses. By aggregating this feedback across the fleet, the server can identify which algorithms are most reliable and which vehicles need updates, creating a continuous improvement loop that enhances overall fleet reliability.
Solution Approach 2:
Vehicles automatically report their system status and performance data to the server without requiring manual intervention. The server then automatically distributes update information to vehicles that need it, enabling the fleet to self-upgrade and maintain high reliability across all vehicles.
2Reliability
If early warning systems use locally limited approaches via car-to-2X communication, then warning can be provided to drivers, but the notice time is very short and does not always provide necessary period for driver to take control
Solution Approach 1:
The server analyzes data from multiple vehicles to predict dangerous situations before they occur. By identifying potential hazards in advance and notifying drivers beforehand, the system provides sufficient time for drivers to prepare and take control when needed, rather than reacting only when danger is immediate.
Solution Approach 2:
The server acts as an intermediary that collects and analyzes data from multiple vehicles, then provides coordinated warnings to multiple vehicles ahead of time. This intermediary approach allows the system to process information centrally and distribute timely warnings to all affected vehicles, extending the notice period beyond what single-vehicle systems can achieve.
3Adaptability or versatility
If multiple identically constructed vehicles use different algorithms for examining feasibility of autonomous driving mode, then vehicles can adapt to different conditions, but the performance of autonomous driving mode is inconsistent across the fleet
Solution Approach 1:
The server evaluates different algorithm parameters and configurations across the fleet, identifying which parameter settings produce the most reliable autonomous driving performance. By standardizing on the best-performing parameters while allowing conditional adaptations, the system maintains both versatility and consistency across the vehicle fleet.
Data Source
AI summary
A driver assistance method by a control device is to receive driving environmental data from identically constructed motor vehicle sensor technologies of a first motor vehicle, as well as from at least one further motor vehicle, which drive on a set road section in an at least partially autonomous driving mode. Each motor vehicle may provide a road section model for each of the motor vehicles with an algorithm recorded in each respective motor vehicle. The control device examines the road section models and/or the algorithms for a feasibility of the at least partially autonomous driving mode for the first motor vehicle, evaluates the algorithms of each motor vehicle based on a comparison of the road models, and sets that algorithm, which satisfies the set feasibility criterion with the highest probability, as the target algorithm. The control device sets the road section as a check section if the set feasibility criterion is not satisfied for the first motor vehicle, and generates and transfers an early warning signal to the further motor vehicles.