I. Introduction to Intelligent Robotic Vacuum Cleaners
There's a sensor at the front that can detect obstacles. If it hits a wall or other obstacle, it will open automatically. Depending on the manufacturer's settings, it will take different routes and schedule cleaning of the area. Due to its simple operation and convenience, it has gradually become popular and a common household appliance for working professionals and modern families.
Today, robotics technology is becoming increasingly mature, so each brand has different research and development directions and special designs, such as: dual vacuum cleaner covers, connected handheld vacuum cleaners, dust boxes that can be washed and wiped, can be scented, or have photocatalytic sterilization functions.
II. Visual Navigation Technology for Intelligent Robotic Vacuum Cleaners
Visual navigation-equipped robotic vacuum cleaners use a monocular solution for positioning. This type of intelligent robotic vacuum cleaner has its visual sensor mounted on top, and through sophisticated algorithms, it can locate itself by perceiving optical images composed of light spots with varying brightness. These optical images appear different from different angles. By continuously collecting image information, the robot can position itself on a self-built map, thus knowing which areas have been scanned and which areas need cleaning.
The advantages of visual navigation are mainly twofold: low cost and wide applicability, and it can leverage big data to address technical challenges. However, visual navigation cleaners also have significant drawbacks. Accuracy is a prerequisite for visual ranging. After the camera acquires environmental information, it must quickly calculate the distance. If data processing and algorithms cannot keep up during this process, the distance data will be inaccurate.
Therefore, from a technical implementation perspective, visual navigation is undoubtedly much more difficult than laser navigation. This is why there are very few robotic vacuum cleaners on the market that fully utilize visual navigation systems; more often, intelligent robotic vacuum cleaners use a fusion solution of laser navigation and visual sensors or other sensors.
III. Global Path Planning and Local Path Planning
When discussing path planning for robotic vacuum cleaners, SLAM (Simultaneous Localization and Mapping) technology must be mentioned. Unsurprisingly, SLAM and path planning are two separate concepts. SLAM is more like a passive skill, providing the robot with maps and location information. For intelligent robotic vacuum cleaners to achieve autonomous movement, they require the cooperation of path planning and SLAM. Without SLAM providing high-quality localization information for path planning, intelligent robotic vacuum cleaners will struggle to achieve proper path planning.
For intelligent robotic vacuum cleaners, path planning involves more than just solving the inherent difficulties of SLAM (Single-Led Aided Mapping). We can simplify path planning as a movement problem from point A to point B. This leads to the question of how to perform global and local path planning.
So, what exactly is global path planning? We can understand it this way: on a static map, the robot calculates the distance from its current point to the target point based solely on the map. There are many algorithms for this approach, the most common being the ASTAR algorithm. In practical applications, this method has a very wide range of uses.
Besides global path planning, another issue is local path planning. In global path planning, the robotic vacuum cleaner has planned a possible walking path. However, reality is not always ideal; many unexpected situations may arise during actual movement. For example, someone might suddenly walk by and block the robot's planned path. In this situation, how can the robotic vacuum cleaner bypass the person without modifying its previously planned path? For the robotic vacuum cleaner, the general walking direction is correct, but if there is an obstacle, it needs to temporarily change its path. This process is called local path planning. Currently, the algorithm traditionally used to solve this problem is EFF, which employs the current dynamic permeability algorithm.
