Autonomous Driving Edge Nodes for Faster Sensor-Based Decisions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current autonomous driving systems face challenges such as increased costs due to independent sensor systems and long service execution delays caused by centralized data processing and decision-making.
Innovation Solution
An autonomous driving system is implemented where sensor data is processed by edge nodes directly connected to the sensors, enabling real-time perception information and decision-making, while a central node delegates state decision functions to these edge nodes to reduce service execution delays.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If various functions of the vehicle develop independently (like chimneys), then each function can be implemented with dedicated sensor systems, but the overall cost of the vehicle increases and data cannot be reused between functions
Solution Approach 1:
The patent implements a centralized sensor system where sensors serve multiple functions simultaneously. For example, the same sensor system is used for both surround view perception and emergency braking assistance, allowing data to be reused across different functions. This universal approach reduces the need for separate dedicated sensor systems for each function, thereby reducing overall system cost while maintaining functional independence through software-level function separation.
Solution Approach 2:
The patent merges previously separate sensor systems into a unified sensor platform. By combining the sensor infrastructure used for surround view systems with the sensor infrastructure needed for emergency braking assistive systems, the patent eliminates redundant hardware. The centralized sensor system processes data that can be shared across multiple functions, reducing overall system complexity and cost while maintaining the ability to independently implement different vehicle functions.
2Device complexity
If various functions of the vehicle are handled in a centralized manner with a central node processing all raw data, then data can be centrally managed, but the service execution delay increases due to hierarchical data reporting
Solution Approach 1:
The patent segments the centralized processing architecture into distributed edge computing nodes. Instead of all raw data being reported to a single central node through hierarchical layers, the system divides processing tasks across multiple edge nodes that are closer to the sensors. Each edge node can independently process data for specific functions, enabling parallel processing and reducing the time delay associated with hierarchical data reporting while maintaining centralized management capabilities.
Solution Approach 2:
The patent introduces a spatial dimension to data processing by deploying edge computing nodes at multiple locations closer to the sensors rather than relying solely on a centralized hierarchical structure. This dimensional change allows data to be processed in parallel across multiple nodes simultaneously, significantly reducing service execution delay while maintaining the benefits of centralized management through coordinated control.
3Extent of automation
If raw data collected by sensors is reported to the central node layer by layer, then centralized decision-making can be achieved, but the control instruction delegation and data reporting process becomes slow
Solution Approach 1:
The patent segments the centralized decision-making process into distributed edge computing nodes that can autonomously process data and generate control instructions for their respective functions. This segmentation allows parallel processing of multiple functions simultaneously, significantly improving service execution speed while maintaining the automated decision-making capability. Each edge node operates autonomously within its domain, eliminating the sequential layer-by-layer reporting bottleneck.
Solution Approach 2:
The patent enables edge computing nodes to perform self-service by autonomously processing sensor data and generating control instructions without requiring constant central node intervention. Each edge node independently executes its designated functions, making real-time decisions based on local data processing. This self-service capability dramatically improves service execution speed while maintaining centralized oversight for coordination and management.
Data Source
Figure 1A~1B
Figure 1C~1D
Figure 2
AI summary
An autonomous driving system and an autonomous vehicle are disclosed. The system includes a plurality of sensor combinations, where each of the plurality of sensor combinations includes at least one sensor; a plurality of edge nodes (120), where the plurality of edge nodes (120) are in a one-to-one correspondence with the plurality of sensor combinations, and a first edge node in the plurality of edge nodes is configured to determine first perception information of a vehicle based on data collected by a first sensor combination corresponding to the first edge node; and a central node (110) connected to the plurality of edge nodes, where the central node (110) is configured to delegate a first state decision function to the first edge node in the plurality of edge nodes (120), so that the first edge node determines that the vehicle enters a first state or determines a first driving policy of the vehicle in the first state by executing the first state decision function. Execution of the first state decision function requires at least the first perception information. This reduces service execution time.