What are the common obstacle avoidance solutions?
Navigation and obstacle avoidance are crucial and challenging issues for robotic vacuum cleaners today. Various obstacles can appear at home, and cleaning cannot generally proceed once the cleaner suddenly gets entangled or jammed. The robot might get stuck in obstacles and stop working, so you must rescue it. There are common obstacle avoidance solutions out there for robotic vacuums, such as LiDAR, 3D structured light, 3D TOF, single camera, etc. Let's introduce them briefly.
|
Solution
|
Pros
|
Cons
|
|
LiDAR
|
1.High precision: Measures distances and object locations accurately, creating detailed maps of the environment. 2.Fast response: Updates environmental information in real-time, quickly adjusting the path. |
1.Difficulty recognizing black or transparent objects: Low reflectivity on these surfaces makes detection challenging. |
|
3D Structured Light
|
1.Accurate depth information: Precisely captures complex shapes and small objects. 2.Good spatial recognition: Suitable for navigating complex indoor environments. |
1.Accuracy decreases when the distance increases. 2.It is easily affected by intense light. |
|
3D ToF (Time of Flight)
|
1.Rapid data acquisition: Provides real-time depth information of objects. 2.Accurate distance measurement: Suitable for quick obstacle avoidance in dynamic environments. |
1.The recognition resolution is relatively low. |
|
Single Camera
|
1.Low cost: Reduces manufacturing costs by using a single camera. 2.Simple and practical: Suitable for basic obstacle avoidance needs and simple home environments. |
1.Recognition limitations: May struggle with complex and variable environments. 2.Need other sensors to detect distance. 3.Dependent on lighting: Low light conditions or reflective surfaces can affect detection. |
What is the Narwal Freo Z Ultra’s solution?
The Narwal Freo Z Ultra features the world’s first TwinAI obstacle avoidance system, equipped with dual RGB cameras and dual AI computing chips. It precisely avoids obstacles in complex environments while maximizing cleaning coverage, so you are not bothered at all during cleaning.
How does the Narwal Freo Z Ultra know obstacles?
First, let's discuss the cameras. The Freo Z Ultra is equipped with two high-definition RGB cameras, each with a resolution of 1600x1200. Generally, one camera captures a planar view (2D model), while binocular visual cameras create a stereoscopic image (3D model), similar to human eyes. This technology enables the robot to perceive an object's distance, size, and material, among other characteristics.
Benefiting from the dual cameras, the robot's field of view can reach an ultra-wide 136 degrees, which is broader than the 90 degrees in others. Transparent or black items, such as plastic bags, gamepads, etc., are easily identified with dual cameras. This capability allows the robot to quickly capture more details in a complex environment and make agile adjustments to its route and cleaning strategy.
Freo Z Ultra's scanning ability is around 1.5 million voxels per second, offering 750 times more precise recognition than the ordinary structured light solution. This technology leads to millimeter-level measurement accuracy and spatial detection, providing visual judgment akin to a human's. It can even recognize items as small as 5 millimeters, while others cannot detect objects smaller than 2 centimeters. Therefore, the Freo Z Ultra can detect more than 120 objects, including transparent, small items, and complex furniture.
We mentioned the importance of lighting for visual solutions. In order to work in the dark, the robot activates its light in low-light conditions
The 3D Vision from Freo Z Ultra's Dual RGB Cameras
How does the Narwal Freo Z Ultra process data?
With so many images, you might wonder, won't it slow down the machine? The answer is no. We have specifically added a chip dedicated to processing obstacle avoidance data. It is called the industry's first dual AI chips technology. It recognizes information faster and analyzes and processes it in real time, so there are no restrictions on cleaning even in more complex home environments.
Narwal Freo Z Ultra utilizes an architecture that includes a main chip and an AI-dedicated chip. The 4-core main CPU chip is responsible for system running and cleaning decisions. In addition, the AI-dedicated chip provides 4 Trillion Computations per Second (4TOPs) of autonomous driving-level computing power, handling visual recognition and obstacle avoidance.
The critical aspect lies in the differentiated intelligent routing and strategy, which adapts to various objects using these technologies.
Step 1. The process begins with the dual cameras capturing and transmitting the scene to the AI chip.
Step 2. This chip analyzes the images to determine the robot's current position and identify obstacles ahead.
Step 3. This information is then passed to the main chip, which formulates a tailored cleaning strategy and dispatches commands to the robot.
Last but not least, privacy is paramount. All personal and environmental data collected by the robot are stored locally on the AI chip, ensuring privacy. Additionally, data uploads are disabled by default to protect user information further.
How does it benefit in the real world?
1.Ultra-Precise Obstacle Avoidance
In daily cleaning, different types of obstacles can be encountered, but the Narwal Freo Z Ultra handles each with ease, for examples:
Cables & Wires: Whether it's charging cables or headphone wires, these often cause problems for typical robot vacuums. The Freo Z Ultra swiftly identifies and avoids them without getting tangled, even when cords are left flat on the floor.
Tiny Toys: The Freo Z Ultra can spot and dodge objects as small as 5mm, like mini Legos or pet toys, with pinpoint precision. It glides effortlessly through cluttered toy areas, navigating smoothly and seamlessly—so quiet and precise, you’ll barely notice it’s there.
Left:Freo Z Ultra VS. Right: Others
*Comparison result from internal test
2.Adaptive Bypass Strategy
The Narwal Freo Z Ultra doesn’t just avoid obstacles—By combining high-resolution cameras and powerful AI processing, the Freo Z Ultra’s adaptive bypass strategy delivers a tailored cleaning experience that guarantees precision, safety, and thoroughness for every surface and obstacle.
it intelligently adapts its cleaning approach for different types of objects, ensuring maximum coverage without compromising safety or cleanliness.
Edge-to-Edge Cleaning for Maximum Coverage
Avoiding cables isn’t enough—true intelligence means using obstacles to its advantage. The Freo Z Ultra can clean along the edge of cables with incredible precision, down to 8mm, maximizing coverage while ensuring no dirt is left behind. Instead of just dodging objects, it works around them, increasing efficiency and leaving your floors spotless, even in tricky spots like table legs.
Safe Distance for Pet Poop
When it comes to more hazardous obstacles, like pet poop, the Freo Z Ultra takes no chances. With its advanced vision system, it detects pet poop in real-time and maintains a safe 15cm distance while cleaning around the area. This ensures your home stays clean and sanitary without the risk of spreading contaminants across your floors.
3.Real-Time Obstacle Avoidance
The Narwal Freo Z Ultra owes its superior obstacle avoidance to its dual AI chips and high-speed processing power, enabling it to react instantly to unexpected obstacles. Thanks to its rapid computation capabilities, the robot scans its surroundings in real-time, making split-second decisions to dodge moving objects like pets or newly dropped items.
Whether a pet suddenly walks into its path or a toy rolls into its cleaning zone, the Freo Z Ultra adjusts its route with incredible precision. This powerful, fast-response system allows it to navigate complex environments effortlessly, ensuring a smooth and uninterrupted clean in even the busiest homes.
