Low-Dimensional Environmental Map Construction Using Orthogonal Projection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Creating a low-dimensional environmental map from high-dimensional sensor data is computationally intensive, making it difficult for small robots with limited resources to process and generate maps in real-time, as existing methods require significant calculation and data handling capabilities.
Innovation Solution
An information processing apparatus that clusters high-dimensional sensor data to construct geometric objects, projects these objects orthogonally onto a reference plane, and performs hidden surface removal using a Z buffer to reduce data and calculation loads, allowing for the creation of low-dimensional environmental maps with height information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If environmental map is created from high-dimensional sensor data using conventional methods, then map accuracy including height information is improved, but calculation amount and data processing load increase enormously
Solution Approach 1:
The sensor data processing is segmented into distinct functional units: data acquisition, clustering, object construction, projection, hidden surface removal, and map construction. Each unit handles a specific aspect of the processing pipeline, allowing for optimized computation at each stage rather than processing all high-dimensional data uniformly.
Solution Approach 2:
The invention extracts only the necessary geometric information (position, size, shape, posture) from the high-dimensional sensor data to construct simplified three-dimensional objects. This extraction process removes redundant data points while preserving the essential spatial characteristics needed for accurate environmental mapping.
2Measurement precision
If high-dimensional sensor data is processed to create environmental map with height information, then map quality is improved, but processing time increases
Solution Approach 1:
The clustering and object construction steps are performed as preliminary actions before the actual map construction. By pre-processing the sensor data into structured three-dimensional objects with defined geometric properties, the subsequent projection and mapping operations can proceed more efficiently without re-processing raw data points.
Solution Approach 2:
The invention creates simplified three-dimensional object models that copy only the essential geometric characteristics of real-world objects. These simplified models can be rapidly projected and processed compared to the original high-dimensional point cloud data, significantly reducing processing time while maintaining map quality.
3Loss of information
If three-dimensional object construction from clustered sensor data is performed, then data representation accuracy is improved, but memory requirements increase
Solution Approach 1:
The three-dimensional objects are constructed with locally optimized geometric properties - each object stores only the specific attributes (position, size, shape, posture) relevant to its representation. This local quality approach ensures accurate data representation for each object while avoiding the storage of redundant global information that would increase memory requirements.
Data Source
AI summary
Provided is an information processing apparatus that efficiently creates a low-dimensional environmental map on the basis of high-dimensional sensor data. The information processing apparatus includes: an object construction unit that clusters sensor data including a high-dimensional point cloud and constructs a high-dimensional object for each cluster; a projection unit that subjects the high-dimensional object to low-dimensional projection; and a map construction unit that constructs a map on a basis of the object subjected to the low-dimensional projection. The object construction unit constructs a high-dimensional object having a size, a shape, and a posture based on geometric information extracted from a set of sensor data included in a cluster.