Semantic Local Mapping With Onboard RGBD-IMU Sensor Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods require a high-performance server or control system to generate a local map for robot navigation, which is cumbersome and limits the robot's ability to operate in environments without pre-obtained maps.
Innovation Solution
A semantic local map generation device using a multi-sensor unit comprising RGBD and IMU sensors, which estimates the robot's pose and generates a semantic local map in real time, enabling the robot to differentiate driveable and undriveable regions without a pre-existing map.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a high-performance server or control system is used to generate a local map, then the map generation accuracy is improved, but the device complexity and cost increase
Solution Approach 1:
The robot performs local map generation autonomously using its own onboard sensors (RGBD camera, IMU, laser range finder) and processing unit, eliminating the need for external high-performance servers. The system self-services by integrating all necessary components within the robot itself, achieving map generation without external infrastructure support.
Solution Approach 2:
The patent extracts the map generation functionality from external servers and implements it directly on the robot using lightweight algorithms optimized for embedded systems. By taking out the complex server-based processing and replacing it with simplified onboard processing, the system reduces device complexity while maintaining functional capability.
2Reliability
If a pre-obtained map is obtained in advance, then the robot navigation reliability is improved, but the loss of time for map acquisition increases
Solution Approach 1:
The system performs preliminary actions by continuously maintaining and updating the local map in real-time as the robot moves, rather than acquiring the complete map in advance. This allows the robot to have navigation-ready map data available immediately for new areas, eliminating the time-consuming pre-mapping phase while ensuring reliability through continuous real-time updates.
Solution Approach 2:
The local map is dynamically updated in real-time as the robot navigates, adapting to new environments continuously. This dynamic approach allows the system to transition from static pre-obtained maps to living, evolving maps that are always current, reducing the time needed for map acquisition while maintaining navigation reliability through up-to-date spatial information.
3Power
If a separate high-performance server is used for map generation, then the processing power is improved, but the ease of operation decreases
Solution Approach 1:
The robot's onboard processing unit performs multiple functions including sensor data acquisition, real-time pose estimation, semantic segmentation, and local map generation. By making the robot self-sufficient with multi-functional onboard processing, the system eliminates the need for separate external servers, simplifying operation while maintaining adequate processing power through integrated architecture.
Solution Approach 2:
The robot serves itself by performing all map generation operations autonomously using its own onboard computer and sensors. This self-service capability eliminates the need for external server infrastructure, making the system easier to operate as it requires no external support while maintaining sufficient processing power through dedicated onboard computing resources.
4Adaptability or versatility
If real-time local map generation is implemented without pre-obtained maps, then the adaptability is improved, but the measurement precision may deteriorate
Solution Approach 1:
The system merges multiple sensing modalities (RGBD camera for color and depth, laser range finder for precise distance measurement, IMU for orientation) to compensate for the challenges of real-time map generation. By combining these diverse sensors with different strengths, the system achieves both adaptability to unknown environments and measurement precision through sensor fusion and cross-validation of data from multiple sources.
Solution Approach 2:
The system performs preliminary pose estimation using IMU data before refining it with visual and range data. This preliminary action provides an initial accurate spatial framework that guides subsequent detailed mapping operations, ensuring measurement precision is maintained even in real-time adaptive environments without pre-obtained maps.
Data Source
AI summary
A semantic local map generation device may include a multi-sensor unit including a RGBD sensor and an inertial measurement unit (IMU) sensor attached to a body of a robot, and a data processing unit operatively connected to the multi-sensor unit and configured for estimating a pose of the robot and a semantic point cloud with respect to a driving region from sensor data obtained from the multi-sensor unit, and generate a semantic local map based on the estimated pose and the estimated semantic point cloud.