Advanced Obstacle Avoidance System behind the Lymow One

With the boom of smart home and AI technologies, automatic lawnmowers are fast becoming the go-to solution for households wanting pristine, healthy lawns. But to perform safely and effectively in the ever-changing conditions of outdoor spaces, smart obstacle avoidance has emerged as a vital consideration for buyers. In this blog, we’ll explore the primary types of sensors used for obstacle detection in today’s lawnmowers, assessing their strengths and weaknesses, and delve into the advanced obstacle avoidance system behind the Lymow One.

A Look at Common Obstacle Avoidance Sensors

Physical Bumper Sensors

When it comes to obstacle avoidance, physical bumper sensors are widely used in many types of robots due to their simplicity and effectiveness, especially robotic lawn mowers. When the mower encounters an object, the bumper triggers a sensor or mechanical switch, causing the machine to stop and change direction.

Pros:

• Simplicity: No complicated electronics required—just a simple mechanical design that reliably detects obstacles.
• Affordability: Both production and maintenance are cost-effective.
• Reliability: Works well in most situations, especially when encountering hard objects.
• Compatibility: It can be easily integrated into existing mower designs.

Cons:

• Passive obstacle avoidance: It only reacts upon actual contact with an obstacle, which may cause minor damage to both the machine and the obstacle.
• Lack of foresight: It cannot detect obstacles in advance, leading to reduced mowing efficiency.
• Limited applicability: It may not effectively identify soft or small obstacles, such as small stones.

 

Ultrasonic Sensors

Ultrasonic sensors play a crucial role in outdoor robots by emitting ultrasonic pulses and measuring the time it takes for the echoes to return. This process allows them to accurately determine the presence and distance of obstacles ahead. With a large Field of View (FOV), multiple ultrasonic sensors strategically positioned at different angles can effectively manage obstacle avoidance for the entire machine.

Pros:

• Cost-Effectiveness: Ultrasonic sensors provide a budget-friendly solution without sacrificing performance.
• User-Friendly Installation: The compact design facilitates easy integration in various locations on the mower.
• Weather Resistance: Built to excel in all weather conditions, ultrasonic sensors ensure reliable operation even at nighttime or in gloomy
• Rapid Response: Quickly identify obstacles and react promptly.

Cons:

• Resolution Limitations: They may struggle to detect very small objects, and their effectiveness decreases with distance.
• Challenges with Soft or Transparent Objects: Items like plastic bags or fine lines may not be effectively detected.
• Environmental sensitivity: Performance can be significantly affected by temperature, humidity, wind speed and other factors.
• Limited detection range: Ultrasonic sensors often have blind spots, typically providing effective detection only within a few meters

ToF (Time of Flight) Sensors

ToF sensors utilize infrared or laser technology to measure the time it takes for light to travel to an object and back, allowing them to calculate the distance to obstacles accurately.

Pros:

• Rapid Response: Provide real-time, precise distance information.
• High Precision: Compared to ultrasonic sensors, ToF sensors offer superior accuracy.
• Compact Design: Their small size makes them easy to integrate into compact devices without adding significant weight.
• Low Power Consumption: Requires minimal energy to operate, helping to extend battery life.

Cons:

• Susceptibility to Strong Light: Intense sunlight or other bright light sources can interfere with sensor performance.
• Higher Cost: Although less expensive than LiDAR, ToF sensors are still more expensive than ultrasonic sensors, with higher resolutions leading to increased costs.
• Calibration Requirements: Regular calibration is necessary to maintain accuracy.

 

LiDAR (Light Detection and Ranging)

LiDAR constructs a 3D map of the surrounding environment by emitting laser beams and receiving reflected signals.

2D LiDAR

Also known as single-line LiDAR, it primarily operates on a horizontal plane, providing high-precision distance information that is widely used in fields such as robotic navigation, security monitoring, industrial automation, and robotic lawnmowers.

Pros:

• Precise Data on Horizontal Planes: Delivers detailed environmental information on plane surfaces.
• Stability: Performs reliably in flat terrains, ensuring consistent operation.

Cons:

• Lack of Vertical Information: Limited adaptability to terrain with significant height variations.
• Higher Cost: More expensive compared to other sensors.
• Large Size: Less suitable for compact designs.

 

3D LiDAR (Light Detection and Ranging)

3D LiDAR utilizes laser beams to measure distances and create 3D maps of the environment. By emitting laser pulses and receiving the reflected signals, it calculates the distance, shape, and location of objects. This technology is extensively applied in autonomous vehicles, drones, terrain mapping, intelligent robots, and automatic lawnmowers.

Pros:

• Detailed Information in 3D Space: Capable of generating comprehensive 3D maps.
• Effective in Complex Terrain: Functions efficiently even in uneven landscapes, adapting well to various environments.

Cons:

• Elevated Cost: Particularly expensive for mechanical rotating LiDAR systems, limiting its use in consumer products.
• Size and Weight: Mechanical rotating LiDAR tends to be bulky and heavy, making it unsuitable for compact designs.
• High Power Consumption: These systems consume significant energy during operation, potentially impacting battery life.
• Susceptibility to Environmental Factors: While less affected by lighting conditions, performance can decline in extreme weather, such as heavy fog or rain.
• Complex Data Processing: The large amounts of point cloud data generated require substantial computational power and efficient algorithms for processing.

Vision Sensors

Vision sensors mimic the function of the human eyes, using cameras to capture and process images for obstacle detection and navigation. This approach aligns with the First Principles.

Binocular Vision

Pros:

• Mimicking the principle of human binocular disparity: Enable depth estimation and provides 3D information.
• Higher Precision: Compared to monocular vision, binocular vision provides superior depth accuracy.

Cons:

• Light Sensitivity: Variations in lighting can impact image quality and, consequently, depth estimation.
• High Computational Demand: Image processing and depth calculation require substantial computational resources.
• Calibration Requirements: Precise calibration of both cameras is essential for accuracy.

 Multi-Camera Vision/360-Degree Vision

Pros:

• Wide Field of View: Capable of monitoring the entire surrounding environment.
• Multiple Perspectives: Enhance understanding of complex scenarios.

Cons:

• Complex Data Processing: Managing data from multiple cameras increases computational load.
• Integration Challenges: Synchronization and calibration of several cameras complicate system integration.
• Elevated Cost: Requires additional cameras and hardware support, increasing overall expenses.

 

Comprehensive Obstacle Avoidance Strategy of the Lymow One Robotic Lawnmower

The Lymow One robotic lawnmower features an advanced multi-layered obstacle avoidance strategy that combines binocular vision sensors, multiple ultrasonic sensors, and physical bumpers. This innovative integration leverages the strengths of each technology to overcome the limitations of any single method, ensuring safe and efficient operation in various environments.

 

The Combination of Semantic Understanding and 3D Perception

From the outset, the design of the Lymow One included extensive research on various sensors, highlighting the importance of visual sensors. Semantic information is one of the most complex data types in a yard, making its perception and processing vital for the future of robotic lawnmowers—just as camera perception is essential for vacuum cleaners.

In contrast:

• Robotic mowers that rely solely on physical bumpers and ultrasonic sensors cannot tackle complex semantic challenges and resort to basic obstacle avoidance.
• Sensors like radar, which provide only distance information, excel at detecting obstacles that are above ground level. However, they struggle with lower ones, such as picnic blankets, hoses, or sprinkler heads, making them "incomplete" in obstacle detection.

Thus, we believe binocular vision sensors are an optimal choice for contemporary yard robots, offering several strengths:

• Practicality: Binocular vision captures rich environmental data, facilitating navigation and obstacle avoidance in complex scenarios, and is commonly used for 3D scene reconstruction.
• Stability: Widely proven in mobile robotics, autonomous vehicles, and drones, binocular vision has demonstrated remarkable stability.
• High Scalability: As a foundational input for multimodal fusion algorithms, vision sensors offer great scalability, enabling applications like semantic perception, boundary detection, and environmental recovery.

Based on this analysis, Lymow One employs binocular vision for semantic understanding and 3D reconstruction of its environment, allowing it to grasp complex yard scenarios from two perspectives:

1. Its powerful perception capabilities grant the mower semantic intelligence, enabling a deeper understanding of the surroundings.
2. 3D reconstruction through binocular vision ensures foundational safety, allowing the mower to navigate obstacles securely even when semantic data is lacking.
3. The combination of these two aspects results in a smart, safe, and high-performance obstacle avoidance system.

Conclusion

This article provides an in-depth overview of the mainstream obstacle avoidance sensors used in smart robotic lawnmowers, highlighting the unique advantages and disadvantages of each.

For Lymow One, we established a vision-centric design approach from the beginning, ultimately selecting a mature and stable binocular vision system. By integrating semantic understanding with 3D perception, we offer a robust obstacle avoidance solution that balances intelligence and safety.


Lymow One on Kickstarter:
https://www.kickstarter.com/projects/lymowone/lymow-one-boundary-free-robot-mower-for-any-terrain-and-size