Integer-Based Occupancy Grid Fusion for Low-Power Robotics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multi-sensor fusion methods for robotics, particularly in navigation and perception, require significant computing power and energy due to the use of floating-point calculations, making them unsuitable for embedded systems and leading to high energy consumption and heat dissipation issues.
Innovation Solution
A method and system for multi-sensor fusion using discrete models and integer calculations to construct a consolidated occupancy grid, reducing the need for floating-point operations and minimizing energy consumption by merging probability classes with integer indices, allowing for efficient perception of material bodies on simple on-board devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If floating-point calculations are used for multi-sensor fusion, then measurement precision and reliability are improved, but energy consumption and device complexity increase significantly
Solution Approach 1:
The patent changes the numerical parameter type from floating-point to integer arithmetic. By representing occupancy probabilities as integer values (0-255) instead of floating-point numbers, the system maintains sufficient precision for occupancy grid calculations while enabling the use of simple integer arithmetic operations that consume significantly less energy and can be executed on microcontrollers without floating-point units.
Solution Approach 2:
The patent employs simplified integer-based probability representations that can be computed and discarded efficiently. By using discrete integer probability classes instead of precise floating-point values, the system trades minimal precision for dramatically reduced computational cost, allowing frequent updates of occupancy grids without prohibitive energy expenditure.
2Measurement precision
If floating-point calculations are used for multi-sensor fusion, then measurement precision is improved, but device complexity and computing resource requirements increase
Solution Approach 1:
The patent transforms the computational parameter representation from floating-point format to integer format. This change enables the use of simple integer arithmetic operations (addition, subtraction, multiplication, division) instead of complex floating-point operations, allowing implementation on simple microcontrollers and embedded systems without requiring sophisticated floating-point processing units.
Solution Approach 2:
The patent substitutes the computational mechanism from floating-point arithmetic to integer arithmetic. This substitution replaces complex computational hardware requirements with simple integer-based processing, making the system compatible with resource-constrained embedded devices while maintaining adequate precision for occupancy grid fusion applications.
3Measurement precision
If floating-point calculations are used for multi-sensor fusion, then measurement precision is improved, but heat dissipation and reliability worsen
Solution Approach 1:
By changing the numerical representation from floating-point to integer values, the patent reduces the computational complexity of each arithmetic operation. This reduction translates to fewer transistor switching events and less electrical resistance heating, thereby decreasing heat generation during occupancy grid calculations and improving thermal management in embedded robotic systems.
Solution Approach 2:
The patent uses simplified integer probability representations that require minimal computational resources to process. This approach reduces the sustained high-power operation needed for floating-point calculations, thereby reducing cumulative heat generation and improving system reliability in temperature-sensitive embedded environments.
4Use of energy by moving object
If integer calculations are used for multi-sensor fusion, then energy consumption and device complexity are reduced, but measurement precision deteriorates
Solution Approach 1:
The patent changes the numerical precision level from floating-point (typically 32 or 64 bits) to integer format (8 bits, values 0-255). This parameter change is sufficient for occupancy probability representations where extreme precision is not required, as the discrete nature of occupancy grids and the probabilistic interpretation allow adequate accuracy with coarser quantization.
Solution Approach 2:
The patent applies integer arithmetic with sufficient precision for the application requirements without over-engineering the precision level. By using 8-bit integer values to represent occupancy probabilities, the system provides more than adequate precision for robotic navigation and obstacle detection while avoiding the excessive energy consumption and hardware complexity of full floating-point arithmetic.
Data Source
Figure 1~3
Figure 4~5D
Figure 6A~6B
AI summary
A method for perceiving physical bodies comprising the following steps: a) acquiring a plurality of measurements of the distance of said physical bodies, from one or more sensors (C1 … CNC); b) applying a sensor inverse model to an occupancy grid in order to determine a probability of occupation of a set of cells of said grid; and c) constructing a consolidated occupancy grid by merging the occupation probabilities estimated during step b); characterised in that each so-called sensor inverse model is a discrete model, associating, with each cell of the corresponding occupancy grid, and for each distance measurement, a class of probability chosen within a given set of finite cardinality and identified by an integer index; and in that said step c) is implemented by means of integer calculations carried out on the indices of the classes of probabilities determined during said step b). A system for perceiving physical bodies suitable for implementing such a method.