Mapping a space using a multi-directional camera
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current robotic devices with limited computing resources face challenges in navigating and mapping three-dimensional spaces efficiently due to the complexity and computational intensity of existing techniques, making them unsuitable for low-cost domestic applications.
Innovation Solution
A robotic device equipped with a monocular multi-directional camera that captures images from various angular positions, determines pose data, estimates depth values, and populates an occupancy map using a volumetric function, enabling navigation and cleaning patterns without the need for expensive sensors like LADAR.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If expensive specialized sensors like LADAR, structured light sensors, or time-of-flight depth cameras are used, then measurement precision and mapping accuracy are improved, but device complexity and cost increase
Solution Approach 1:
The patent creates a virtual copy of the depth sensing capability through computational methods. Instead of using physical depth sensors like LADAR, the system captures multiple 2D images from different angles and synthesizes a 3D depth map through image processing algorithms, effectively copying the depth measurement function using only a standard camera
Solution Approach 2:
The patent replaces the mechanical/optical depth sensing system (LADAR, structured light, time-of-flight cameras) with a computational image processing system. The depth information is derived through algorithmic processing of 2D images rather than direct physical measurement, substituting mechanical depth sensing with computational reconstruction
2Productivity
If complex mapping algorithms with substantial computational resources are used, then productivity and mapping speed are improved, but ease of operation and applicability to embedded devices worsen
Solution Approach 1:
The patent divides the mapping process into distinct segments: capturing images at multiple angular positions, detecting features in each image, matching features across images to determine camera poses, and finally generating the depth map. This segmentation allows the complex task to be performed in manageable steps suitable for embedded processing
Solution Approach 2:
The patent uses a monocular camera to capture images at multiple angular positions (excessive action in terms of number of images) to compensate for the limited information from a single viewpoint. By taking more images than a single depth sensor would require, the system achieves adequate depth measurement capability through computational redundancy
3Device complexity
If a monocular multi-directional camera is used instead of stereo or depth sensors, then device complexity and cost are reduced, but measurement precision and depth estimation accuracy worsen
Solution Approach 1:
The patent transitions from 2D image data to 3D depth information by capturing images at multiple angular positions around the robot. The angular position dimension is used to reconstruct depth, converting a series of 2D views into a 3D depth map through geometric relationships between viewpoints
Solution Approach 2:
The patent merges multiple 2D images captured at different angular positions with their corresponding camera pose information to create a unified 3D depth map. By combining information from multiple viewpoints and integrating it through volumetric functions, the system achieves depth estimation accuracy comparable to dedicated depth sensors
Data Source
AI summary
Examples described herein relate to mapping a space using a multi-directional camera. This mapping may be performed with a robotic device comprising a monocular multi-directional camera device and at least one movement actuator. The mapping may generate an occupancy map to determine navigable portions of the space. A robotic device movement around a point in a plane of movement may be instructed using the at least one movement actuator. Using the monocular multi-directional camera device, a sequence of images are obtained (610) at different angular positions during the instructed movement. Pose data is determined (620) from the sequence of images. The pose data is determined using features detected within the sequence of images. Depth values are then estimated (630) by evaluating a volumetric function of the sequence of images and the pose data. The depth values are processed (640) to populate the occupancy map for the space.