UAV Sensor Confidence Fusion for Autonomous Corrective Navigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous flight for UAVs faces challenges in dynamic environments due to the lack of defined corrective actions for various unpredictable situations, such as sensor conflicts, weather beyond capabilities, communication losses, and untrusted inputs, which existing AI systems struggle to address effectively.
Innovation Solution
A system and method that utilize a plurality of sensors with specified thresholds, a control unit to determine confidence levels, prioritize inputs, generate combined inputs, and decide on mission tasks, and a database to store these inputs and actions, incorporating trusted certificates and historical data to validate and prioritize sensor measurements and take corrective actions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple sensors are used to improve measurement reliability in dynamic environments, then the system can detect more situations, but sensor conflicts and untrusted inputs increase decision-making complexity
Solution Approach 1:
The system changes the parameter of trust by assigning dynamic confidence levels to sensor inputs based on environmental context, sensor performance history, and data consistency. This allows the system to weight sensor inputs differently in different situations, resolving conflicts without requiring complex manual decision-making rules for every possible sensor combination.
Solution Approach 2:
The control unit acts as an intermediary that mediates between multiple sensors and the mission task decision. It introduces a confidence level assessment mechanism that evaluates the reliability of each sensor input and combines them systematically, rather than directly processing raw sensor data through complex decision logic.
2Adaptability or versatility
If the system processes all sensor inputs equally to ensure comprehensive decision-making, then all sensor data is considered, but the system cannot effectively prioritize critical information in dynamic situations
Solution Approach 1:
The system dynamically adjusts the priority and weight of sensor inputs based on changing environmental conditions, mission context, and sensor performance. Confidence levels are not static but change in real-time, allowing the system to adaptively prioritize critical information while maintaining comprehensive consideration of all sensors.
Solution Approach 2:
Different sensor inputs receive different levels of processing and confidence weighting based on their local quality and relevance to the current situation. Critical sensors in relevant environmental conditions receive higher confidence weights, while less relevant or degraded sensors receive lower weights, optimizing decision-making efficiency.
3Reliability
If the system requires high confidence levels for all decisions to ensure safety, then decision accuracy improves, but the system may fail to act in time for critical situations
Solution Approach 1:
The system allows for partial action by enabling mission tasks to proceed with sufficient but not necessarily maximum confidence levels. In critical situations, the system can take excessive action by overriding normal confidence thresholds when multiple sensors indicate urgent conditions, balancing safety with timely response.
Solution Approach 2:
The system performs preliminary confidence assessment and sensor validation in advance, building a foundation of trusted inputs before critical decisions are needed. This preliminary processing reduces the confidence threshold gap during time-critical situations, allowing faster decision-making without sacrificing accuracy.
4Ease of manufacture
If the system uses a simple threshold-based sensor validation to reduce processing complexity, then implementation is easier, but the system cannot handle conflicting sensor data or untrusted inputs effectively
Solution Approach 1:
The sensor validation process is segmented into multiple independent confidence level assessments, one for each sensor input. This modular approach maintains implementation simplicity while allowing each sensor to be evaluated on its own merits and combined systematically, handling conflicts without requiring complex overall validation logic.
Data Source
AI summary
An autonomous vehicle system includes a body and a plurality of sensors coupled to the body and configured to generate a plurality of sensor measurements corresponding to the plurality of sensors. The system also includes a control unit configured to: receive inputs from a plurality of sources wherein the plurality sources comprise the plurality of sensors, the inputs comprise the plurality of sensor measurements; determine a confidence level of each input based on other inputs; prioritize, based on the confidence level associated with each input, the inputs; generate, based on the prioritization of the inputs and the confidence level, a combined input with a combined confidence level; and determine, based on the combined input and the combined confidence level, a mission task to be performed.