Smart Lawnmower Mowing Policy Using Real-to-Sim Training
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current robotic smart lawnmowers are inefficient and ineffective in landscaping and groundskeeping due to reliance on randomness and persistence, failing to penetrate the market effectively.
Innovation Solution
A smart lawnmower that utilizes a real-to-sim training phase to construct a simulated environment based on semantic information, including location signalization, and applies a mowing policy to control the cutting and drive subsystems in a sim-to-real mowing phase, ensuring efficient and automated lawn mowing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If traditional random navigation is used for autonomous mowing, then the lawnmower can operate autonomously without human intervention, but the mowing efficiency and effectiveness are insufficient
Solution Approach 1:
The patent creates a simulated environment that copies the real-world lawn geometry and features. The lawnmower is trained in this virtual replica to learn optimal mowing policies without risking damage in the real environment. This allows efficient learning of productive mowing patterns while maintaining autonomous operation.
Solution Approach 2:
The system performs preliminary training in a simulated environment before deploying the lawnmower to real lawns. The mowing policy is pre-optimized through virtual training, allowing the lawnmower to start with proven efficient behavior rather than random navigation, thus improving productivity from the first real-world operation.
2Productivity
If simulated environment training is implemented, then mowing policy can be optimized for efficiency, but system complexity increases due to real-to-sim and sim-to-real phases
Solution Approach 1:
Instead of complex sensor arrays and processing systems, the patent creates a simplified copy of the real environment in simulation. The geometry and features are replicated virtually, allowing policy training without adding physical complexity to the lawnmower hardware while still achieving optimized productivity.
Solution Approach 2:
The patent replaces complex mechanical navigation and decision-making systems with a software-based simulated environment and policy transfer approach. The training and optimization occur in silico, substituting physical complexity with computational processes that can be executed efficiently.
3Ease of manufacture
If randomness and persistence are used for mowing, then deployment is simple, but the approach fails to penetrate the landscaping market due to inefficiency
Solution Approach 1:
The patent maintains deployment simplicity by using the same hardware platform but adds a virtual environment copy for training. This allows the system to achieve market-competitive effectiveness through software-based policy optimization while keeping the physical deployment process relatively simple.
Solution Approach 2:
The system performs preliminary policy optimization in simulation before real-world deployment. This pre-training step ensures the lawnmower achieves effective productivity before deployment, making the approach competitive in the landscaping market while maintaining relatively simple deployment procedures.
Data Source
AI summary
A smart lawnmower and system and method for use of the same are disclosed. In one embodiment of the smart lawnmower, in a real world-to-simulated world (“real-to-sim”) training phase, the smart lawnmower constructs a simulated environment corresponding to a mowing-relevant portion of a real-world environment relative to semantic information, which may include received location signalization at an antenna In a simulated world-to-real world (“sim-to-real”) mowing phase, a mowing policy is applied to control the cutting subsystem and the drive subsystem in response to the semantic information, which may include the location signalization. In each of the real-to-sim training phase and the sim-to-real mowing phase, the smart lawnmower may provide a user interface including the simulated environment. Further, in the sim-to-real mowing phase, the smart lawnmower may synchronize the real world and the simulated world.