Sensor-Independent Object Interface for Reliable Automated Driving
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In automated driving, existing systems face challenges in balancing false positive and false negative reactions due to erroneous environment perception, particularly in dynamic scenarios where the consistency of sensor measurements is critical for reliable object detection and system reactions like emergency braking.
Innovation Solution
A method and system that calculate continuous sensor-independent existence probabilities using detection probabilities from multiple sensors, accounting for sensor redundancy, visibility, and environmental conditions to assess object reliability, providing a vector representation for decision-making in driver assistance and autonomous driving systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If a single sensor confirms an object, then the system can react quickly to potential hazards, but the reliability of object detection decreases and false positive reactions increase
Solution Approach 1:
The patent segments the sensor confirmation process into multiple independent evaluation dimensions (detection probability, existence probability, measurement consistency) rather than treating object confirmation as a single binary decision. This allows the system to evaluate each dimension separately and combine them for a comprehensive reliability assessment.
Solution Approach 2:
The patent transitions from a single-dimension confirmation approach (one sensor yes/no) to a multi-dimensional probabilistic framework. By introducing separate probability dimensions for detection and existence, and adding temporal consistency evaluation, the system achieves more nuanced reliability assessment without sacrificing reaction speed.
2Reliability
If multiple sensors must confirm an object, then the reliability of object detection increases, but the system may miss real hazards and false negative reactions increase
Solution Approach 1:
The patent changes the evaluation parameters from binary confirmation (confirmed/not confirmed) to continuous probability values (detection probability, existence probability). This allows for graded assessment of object reliability, enabling the system to distinguish between highly reliable detections and uncertain detections without applying a rigid multiple-confirmation rule.
Solution Approach 2:
The patent introduces dynamic temporal evaluation by assessing measurement consistency over time. Rather than statically requiring multiple sensor confirmations, the system dynamically evaluates whether measurements remain consistent across time cycles, allowing flexible adaptation to changing sensor availability and object characteristics.
3Measurement precision
If sensor-specific processing is used, then the accuracy of individual sensor measurements is maximized, but the complexity of the system increases and sensor-independent planning becomes difficult
Solution Approach 1:
The patent introduces probability vectors as an intermediary representation between sensor-specific measurements and sensor-independent planning. The tracking unit converts diverse sensor outputs into a standardized probabilistic format, which the planning unit can then process without needing to understand sensor-specific characteristics. This intermediary layer decouples the complexity of sensor processing from the planning logic.
Solution Approach 2:
The patent transforms sensor-specific measurement data into sensor-independent probability parameters. By converting raw sensor measurements into standardized detection and existence probabilities, the system maintains the precision benefits of sensor-specific processing while presenting a unified, sensor-agnostic interface to the planning unit.
4Adaptability or versatility
If continuous probability values are calculated, then the flexibility in setting thresholds for different criticality levels is improved, but the computational complexity increases
Solution Approach 1:
The patent segments the probability calculation into distinct, modular components (detection probability calculation, existence probability calculation, consistency evaluation). Each component handles a specific aspect of the assessment, making the overall computational process more manageable and easier to optimize than a monolithic probability calculation system.
Data Source
AI summary
A system for a reliability of objects for a driver assistance or automated driving of a vehicle includes a plurality of sensors that include one or more sensor modalities for providing sensor data for the objects. An electronic tracking unit is configured to receive the sensor data to determine a detection probability (p_D) for each of the plurality of sensors for each of the objects, to determine an existence probability (p_ex) for each of the plurality of sensors for each of the objects, and to provide vectors for each of the objects based on the existence probability (p_ex) for each contributing one of the plurality of sensors for the specific object. The vectors are provided by the electronic tracking unit for display as an object interface on a display device. The vectors are independent from the sensor data from the plurality of sensors.