Adaptive Multi-Scale Perception for UAV Landing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current autonomous landing systems for UAVs face inefficiencies due to limitations in perception range and resolution of sensors, leading to unnecessary data storage, communication bandwidth, and processing power waste, as well as inability to robustly handle scale changes and appearance variations, particularly in distinguishing soft ground surfaces from load-bearing surfaces and small foreign objects.
Innovation Solution
An adaptive multi-scale perception system that dynamically selects sensor devices and processing parameters based on altitude, distance, and weather conditions, using a combination of long-range, medium-range, and short-range sensors, including LIDAR and camera systems, to efficiently process sensor data and enhance landing zone detection and object recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the same sensors are used throughout the perception process, then the system can maintain consistent sensor operation, but perception range and resolution are limited and storage space, communication bandwidth, and processing power are wasted
Solution Approach 1:
The system dynamically selects and switches between different sensor devices based on the current operational phase and requirements. During approach phase, long-range sensors are activated; during landing phase, short-range sensors take over. This dynamic sensor selection optimizes measurement precision for each phase while avoiding unnecessary processing of data from inactive sensors, thus reducing processing power waste.
Solution Approach 2:
The perception process is segmented into distinct phases (approach, landing, etc.), and different sensor devices are assigned to different phases. This segmentation allows the system to use appropriate sensors for each phase, improving perception resolution when needed while conserving processing resources by not continuously operating all sensors.
2Reliability
If conventional systems use the same parameters in perception algorithm regardless of altitude or scale, then the system is simple to implement, but the algorithms are not robust to scale change and appearance variations
Solution Approach 1:
The system dynamically adjusts perception algorithm parameters based on altitude and scale changes. As the UAV transitions from high altitude to low altitude, parameters such as region segmentation thresholds and semantic classification criteria are automatically adjusted. This dynamic parameter adaptation enhances algorithm robustness to scale changes and appearance variations.
Solution Approach 2:
The perception algorithm uses different parameters for different operational phases and scales. For example, at high altitude, coarser segmentation parameters are used, while at low altitude, finer segmentation parameters are applied. This parameter change strategy maintains algorithm robustness across varying conditions without requiring a completely different algorithm for each scenario.
3Measurement precision
If a single LIDAR approach is used, then the system is simple in design, but it cannot discriminate soft ground surface over load bearing surface or detect small foreign objects
Solution Approach 1:
The system merges multiple sensor types (LIDAR, cameras, other sensors) into an integrated sensor suite. Different sensors complement each other's capabilities - LIDAR provides depth information, cameras provide texture and color information. This combination enables discrimination of soft ground from load-bearing surfaces and detection of small foreign objects that would be invisible to a single LIDAR system.
Solution Approach 2:
The multi-sensor system provides universal detection capability across different object types and conditions. The same sensor suite can detect both large terrain features and small foreign objects, discriminate between different ground surface types, and operate across various altitudes and lighting conditions, making the system highly versatile.
4Loss of information
If all sensor data is processed and stored at full resolution, then complete information is available, but storage space, communication bandwidth, and processing power are unnecessarily consumed
Solution Approach 1:
The system extracts and processes only the most relevant sensor data at full resolution. During approach phase, only long-range sensor data is processed in detail; during landing phase, short-range sensor data becomes the focus. Less critical data is processed at lower resolution or selectively, reducing storage space consumption while maintaining information completeness for decision-making.
Solution Approach 2:
The system applies partial processing to sensor data based on operational needs. Not all sensor data is processed at maximum resolution at all times - only the data relevant to the current phase and detection goals receives full processing. This partial action approach maintains necessary information completeness while significantly reducing storage and processing requirements.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables efficient and selective storage and processing of sensor data, providing robust perception algorithms for landing zone detection and object recognition, improving the UAV's ability to autonomously select suitable landing sites even in varying conditions.
Implementation Method 1
current art on autonomous landing and object identification in autonomous vehicles has focused on three-dimension (3D) sensors, e.g., LIght Detection And Ranging scanners (LIDAR)
Data Source
AI summary
A system and method of adaptive multi-scale perception of a target for an aircraft, includes receiving sensor signals indicative of aircraft information with respect to the target; selecting one or more sensor devices in response to the receiving of the aircraft information; receiving sensor information from the one or more sensor devices indicative of the target; dynamically selecting sensor processing parameters for the one or more sensor devices; and analyzing the sensor information from the selected sensor devices and using the selected sensor processing parameters.