Dynamic Parameter Server for Reboot-Free ADS Tuning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous driving vehicles face challenges in optimizing default parameters for various high-definition maps and regions, leading to suboptimal performance due to the need for manual rebooting and limitations in detecting certain physical conditions like gentle slopes, which can disrupt normal operation.
Innovation Solution
A dynamic parameter server is introduced to update autonomous driving system parameters in real-time without rebooting, using a configuration file that maps parameters to specific physical conditions, allowing the vehicle to adjust settings based on geographic and other selection factors like map IDs, road IDs, and GPS barriers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If default parameters are manually modified by a user, then the parameters can be customized for specific HD maps or regions, but the operation becomes time-consuming and requires a reboot of software modules which interrupts normal operation
Solution Approach 1:
The system dynamically updates parameters during runtime based on geographic location and HD map data, eliminating the need for manual reconfiguration and reboot. The parameter server continuously monitors vehicle location and automatically adjusts parameters according to the current region's characteristics, making the system adaptive without operational interruption.
Solution Approach 2:
The autonomous driving system performs self-configuration by automatically retrieving and applying appropriate parameters for the current geographic region. The parameter server autonomously matches the vehicle's location with corresponding HD map data and updates parameters without human intervention, eliminating manual configuration time and reboot requirements.
2Productivity
If default parameters are programmatically modified in real-time based on driving conditions, then the operation continuity is maintained, but the parameters may not always be ideal for the HD map and the system cannot detect certain physical conditions like gentle slopes
Solution Approach 1:
The parameter server acts as an intermediary between the autonomous driving system and the configuration files. It retrieves pre-defined optimal parameters from configuration files based on geographic location and HD map data, then applies these parameters to the system. This intermediary approach ensures both real-time operation continuity and parameter optimality by bridging the gap between runtime operations and pre-analyzed optimal settings.
Solution Approach 2:
Optimal parameters for different geographic regions and HD maps are pre-configured in configuration files before runtime. The parameter server retrieves these pre-prepared parameters based on the vehicle's current location, ensuring that optimal settings are immediately available without requiring real-time calculation or manual adjustment, thus maintaining both continuity and optimality.
3Ease of operation
If the system uses fixed default parameters for all regions, then the system operation is simple and stable, but the performance is suboptimal for different HD maps and regions
Solution Approach 1:
The system implements local quality by applying different parameter sets tailored to specific geographic regions and HD maps. The parameter server determines the vehicle's current location and retrieves region-specific parameters from configuration files, ensuring that each local environment receives optimized parameters rather than a one-size-fits-all approach, thereby improving navigation performance while maintaining regional simplicity.
4Measurement precision
If the system requires manual user input to modify parameters, then the parameter accuracy can be high, but the operation complexity and time consumption increase significantly
Solution Approach 1:
The system performs automatic parameter configuration by retrieving and applying appropriate parameters based on geographic location and HD map data without requiring manual user input. The parameter server autonomously completes the entire parameter adjustment process, maintaining high accuracy through pre-configured optimal values while eliminating operational complexity for the user.
Data Source
AI summary
According to one embodiment, a dynamic parameter server is provided in an ADV to update parameters of an autonomous driving system (ADS) of the ADV in real time without requiring the reboot of the ADS. The dynamic parameter server can obtain new parameters from a configuration file created by users based on their experiences and expectations. Each new parameter is mapped to certain physical conditions. When the ADV encounters the physical conditions mapped to a particular parameter, the dynamic parameter server can broadcast the new parameters to the ADS, which can use the new parameters to control the ADV. The physical conditions can be used as selection factors for the dynamic parameter to determine which ADS parameter to update.