Nov 8, 2024 By Lily
Picture this: while commuting home from work, your smartphone unexpectedly fails as a navigation device. Can you still make your way home? GNSS (Global Navigation Satellite System) technology may seem unassuming, but it’s essential for locating ourselves, navigating, and exploring our surroundings. Its accuracy and convenience are ingrained in our daily lives.
Inspired by this, RTK (Real-Time Kinematic), which relies on GNSS satellite signals as its primary source of positioning data has become widely adopted in robotic lawn mowers. This technology allows these mowers to achieve centimeter-level precision, transforming the lawn care process. However, can today’s RTK positioning systems be as "smart" as the GNSS technology used in smartphones? Can they effectively help robotic mowers accurately pinpoint their locations and make lawn maintenance more efficient? Let’s take a closer look.
RTK, or Real-Time Kinematics, is a carrier-phase differential positioning technology that can achieve centimeter-level accuracy, setting it apart from other positioning systems. The reason RTK can achieve real-time centimeter-level positioning accuracy lies in three main factors:
1.1.1 It uses carrier-phase measurements as the distance information, which significantly improves accuracy by 2 to 3 orders of magnitude compared to traditional pseudorange measurements.
1.1.2 By performing differential calculations with data from nearby base stations, various systematic errors can be effectively eliminated.
1.1.3 Although carrier-phase measurements are highly accurate, the uncertainty of the initial phase can lead to the "integer ambiguity" problem, which means not knowing how many complete cycles of the carrier wave are included during signal transmission. However, the RTK algorithm can accurately determine the number of complete cycles, allowing it to achieve centimeter-level positioning accuracy.
In summary, RTK carrier-phase differential positioning combines high-precision carrier-phase observations, differential corrections, integer ambiguity resolution, and suppression of multipath effects to enable centimeter-level high-precision positioning. To illustrate, if the previous accuracy was the length of two Cadillacs, the current accuracy is comparable to the size of a fingernail—exceptionally precise for a backyard. Consequently, robotic lawn mowers can now accurately trim designated areas, similar to contemporary robotic vacuum cleaners.
Additionally, the differential signals between the base station and the rover can be calculated in real-time to provide correction information, and high-frequency data transmission (typically between 1 Hz and 20 Hz) allows for rapid updates to positioning, resulting in RTK systems having a very fast response time, and achieving millisecond-level accuracy. Furthermore, the wide coverage of differential signals helps effectively correct errors caused by multipath effects and environmental interference (such as buildings and trees), reducing the impact of local conditions on the overall positioning process.
However, this does not mean that RTK technology is foolproof. The high dependency of RTK systems on satellite signals from the base station and rover makes them vulnerable to positioning errors, especially in areas with insufficient base station coverage or when there are obstructions like trees. Furthermore, the setup and operation of RTK systems can be quite complex, requiring specialized knowledge and skills for configuration and maintenance, which can hinder the widespread adoption of this technology.
Nevertheless, this technology remains the most widely used positioning method for robotic lawn mowers, while tech companies continue to explore further optimization options, such as the following methods.
As another widely utilized positioning technology, VSLAM (Visual Simultaneous Localization and Mapping) is increasingly being explored by many tech companies due to its low cost and rapid adaptability to environmental changes.
VSLAM is a technique that utilizes visual information for environmental localization and map construction. It integrates computer vision, robotics, and sensor data, enabling real-time map building while accurately determining its position in unknown environments. Typically, VSLAM employs various types of cameras—such as monocular, stereo, or RGB-D cameras—to capture image data from the surroundings, extracting key features like corners and edges from these images. It utilizes image processing algorithms, such as SIFT and ORB, for matching these features and subsequently calculates the robot’s pose—its position and orientation in space—allowing it to gradually construct either a 2D or 3D environmental map.
It may seem intricate, but the core idea is simple: the robot uses its cameras to observe the environment while moving. As it navigates, it gathers visual data to create an accurate real-time map of its surroundings. This mapping enables the robot to make informed decisions about its next actions based on the environment's layout. This technology can be found in various mobile robots, including drones, robotic vacuums, and autonomous vehicles.
However, VSLAM has its drawbacks. The technology requires substantial computational power for real-time processing of large image datasets, demanding efficient algorithms and strong hardware support. Unlike RTK technology, VSLAM is also vulnerable to environmental factors such as lighting changes, colors, and scenery. Rain, snow, and dust can cause inaccuracies in map building, which can hinder the robot's navigation. Additionally, since it depends solely on the mounted cameras, its operational range is somewhat limited.
Lidar SLAM is a real-time localization and mapping method that utilizes laser radar technology. It can produce real-time 2D or 3D maps, typically represented as three-dimensional point clouds or mesh models. You may have noticed that this method is similar to VSLAM, as both involve environmental mapping. However, unlike VSLAM, where the robot uses cameras to capture the surrounding environment, Lidar SLAM operates based on a Lidar sensor mounted on the robot. Lidar sensors measure distances by emitting laser pulses and receiving the reflected signals, generating high-precision point cloud data of the environment.
As the robot moves through its surroundings, it continuously captures point cloud data from each scan to create a map of the current environment, thereby assisting the robot in achieving better localization.
Compared to VSLAM, Lidar SLAM delivers higher-resolution environmental data, especially in complex settings such as city streets and indoor environments, where it achieves greater accuracy than the camera sensors used in VSLAM. Furthermore, since Lidar utilizes lasers for distance measurement, it remains unaffected by lighting conditions, performing reliably in both dim and extremely bright environments. In summary, Lidar SLAM excels in localization capabilities and demonstrates strong resistance to environmental interference.
So, what are the downsides? Well, it mainly comes down to expense. The large amounts of data from the Lidar sensor require significant processing power and resources, which can complicate software development and drive up costs. In comparison, VSLAM typically relies on established visual algorithms, and the abundance of visual sensors provides many open-source software choices. Because of this, a Lidar SLAM system—including the Lidar sensor, computing units, and software—can be several times to many times more expensive than a VSLAM setup. Therefore, VSLAM is usually more accessible for manufacturers and everyday users.
UWB (Ultra-Wideband) positioning is a technology that uses ultra-wide frequency signals for relative positioning and distance measurement. It is primarily applied in indoor positioning scenarios, achieving sub-meter or centimeter-level positioning accuracy. UWB has strong penetration capabilities, effectively passing through walls and other obstacles while maintaining good performance in complex indoor environments, such as asset tracking, personnel localization, and navigation. A particularly notable feature of UWB is its ability to support the positioning and communication of multiple devices simultaneously. Thanks to these advantages, UWB has also started to be used in outdoor robot mowers.
However, its effective communication range is typically quite short, usually between tens and a hundred meters. This limitation may require additional base stations or relay equipment to be deployed for larger lawn areas, which could be restrictive. Additionally, UWB positioning requires the deployment of multiple base stations or anchor devices within the monitored area to achieve accurate positioning, adding complexity and cost to the setup and maintenance. In summary, while it can be effective indoors, there are numerous limitations when used outdoors.
At this point, you might be feeling a bit confused: it seems like none of these solutions are truly reliable?
Through the analysis of the advantages and disadvantages of the various popular positioning technologies mentioned above, it is clear that each technology has its drawbacks. As a result, some lawn mowing robotics companies choose to combine multiple technologies to achieve the best solution. Among the options discussed, RTK technology undoubtedly plays a central role due to its most significant advantages and least drawbacks. We simply need to build on this foundation and incorporate other technologies to maximize the positioning capabilities of robot mowers.
First, let’s consider whether combining RTK and UWB can solve the issues. With RTK, as long as the problems of base station coverage for receiving GNSS signals and signal interference are addressed, the positioning capabilities of RTK should not be in doubt. The base station issue is manageable; as long as there are enough GNSS satellites available in the area, we can set a sufficient number of base stations based on the needs of the lawn mower to address potential signal coverage problems across large areas of grass. But can UWB handle the signal interference issue? That's a bit uncertain.
Second, UWB positioning requires multiple base stations or anchor points to be deployed within the operating area to achieve accurate positioning. This could involve high setup costs and time in large lawn areas or complex environments, explaining why it is more suitable for indoor positioning. Secondly, while UWB signals have a certain degree of penetration ability, they can still be obstructed or attenuated in certain situations (such as dense vegetation or buildings), affecting positioning performance.
What happens when RTK positioning technology is combined with laser positioning technology? From earlier discussions, we know that RTK positioning offers centimeter-level accuracy. Mature technologies like Lidar SLAM also have excellent precision and resistance to interference, meaning their combination could potentially achieve a positioning accuracy where 1 + 1 > 2. However, the technical and maintenance costs associated with such results are prohibitively high, making it nearly impossible for the average consumer to afford. Thus, we will have to hold off for now.
So, can combining RTK and VSLAM create a visual fusion positioning system that ultimately resolves the issue? From a technical perspective, VSLAM addresses RTK's unavoidable signal interference caused by obstacles in the robot's path, as it can continuously build a map to assist the machine in obstacle avoidance, thereby reducing signal interference. From a cost perspective, VSLAM uses mature visual algorithms that have been widely adopted across various sectors, significantly lowering costs. Therefore, both issues are resolved! Meanwhile, RTK can effectively mitigate the shortcomings of VSLAM technology, protecting against potential environmental interference caused by rain, snow, or dust, thus maximizing the positioning outcomes achieved by their combination.
We can take the EFLS 2.0 used in the Navimow robotic lawn mower as an example to specifically examine the application of this technology.
EFLS, or Exact Fusion Locating System, requires a minimum of 10 satellites for stable operation. It combines satellite positioning with signals from multiple sensors to calculate the robot mower's real-time positioning, achieving centimeter-level accuracy. The satellite signals for EFLS are sourced from RTK positioning, with both the Navimow robot mower and an antenna receiving these signals. The charging station facilitates the transfer of signals from the antenna to the mower, enabling synchronous calculations that reduce positioning errors.
However, satellite signals can be obstructed during adverse weather conditions, such as rain or snow. To maintain accurate positioning in such situations, visual landmarks is integrated into the system. In areas with poor satellite coverage, the camera steps in to provide the robot with more dependable positioning. The visual mapping process consists of three key steps:
Step 1: Visual Data Collection
During its initial mapping and the first complete mowing cycle, the mower collects visual data.
Step 2: Visual Map Optimization
Upon returning to the charging station, the mower automatically optimizes the visual map. This optimization takes several hours and will pause if the mower leaves the station.
Step 3: Visual Localization
When RTK signals are weak during mowing operations, the robot mower compares real-time visual features against the visual map data to calculate its location, ensuring uninterrupted mowing.
Moreover, with the help of Visionfence, the robot can quickly respond to obstacles in front of it within a short time, ensuring an efficient mowing process.
Now, let’s enjoy the enhanced efficiency and exceptional lawn maintenance experience brought about by the combination of both RTK technologies in the Navimow robotic lawn mower.
Clearly, the role of positioning technology in lawn maintenance cannot be overstated. Drawing from what we’ve learned, you should now recognize the advantages and disadvantages of various approaches. Despite potential advancements, the integration of RTK and VSLAM stands out as the best current approach. The Navimow robotic lawn mower exemplifies this effective pairing, so be sure to take a closer look!
