AGV obstacle avoidance topic | Detailed explanation of two laser radar obstacle avoidance solutions

程宇

AGV obstacle avoidance topic | Detailed explanation of two laser radar obstacle avoidance solutions [Copy link]

In 2019, the hot topic of warehouse AGV (Automated Guided Vehicle, AGV for short) development was from "shelf to person" to "cargo box to person". In 2020, affected by the national epidemic, companies developed new application areas such as medical logistics AGV. However, no matter how the AGV industry changes and upgrades, it is inseparable from the small but precise subject of "obstacle avoidance".

Figure 1 Warehouse AGV picture

01

Obstacle avoidance method of laser radar in AGV application

There are many different methods for AGV obstacle avoidance on the market, each with its own advantages and disadvantages. This article mainly introduces two main methods of applying Benewake LiDAR to AGV obstacle avoidance: single-point directional obstacle avoidance and solid-state 3D array directional obstacle avoidance.

Click to view full image

Single point directional obstacle avoidance

The Benewake TF series single-point laser radar obstacle avoidance module is based on ToF (Time of Flight) laser ranging technology, which is used for obstacle detection and avoidance at a close distance of 0.1m-12m. The principle and usage are relatively simple. The radar outputs the distance value of the nearest obstacle, thereby guiding the system to brake or decelerate. It is a low-cost, preferred and reliable solution for mobile robots/AGVs to detect short-range low obstacles.

The main features are as follows

High measurement accuracy and stable perception

Small field of view, good collimation, avoid false triggering

100-1000Hz data refresh rate, sensitive obstacle avoidance

Multiple communication interfaces are optional, plug and play

Very low power consumption, light weight, easy to integrate

No echo interference or ground inequality false alarms

Figure 2 Schematic diagram of single-point directional obstacle avoidance laser radar

Figure 3 Distance measurement display of the host computer (there is an obstacle in the circle)

Solid-state 3D array directional obstacle avoidance

3D solid-state laser radar is also a directional obstacle avoidance method, which is different from single-line and multi-line scanning obstacle avoidance methods. The single-line and multi-line rotating radars we are familiar with are installed on the top of the AGV to achieve 360° or specific angle obstacle avoidance, while CE30 detects obstacle information within a fixed line of sight by emitting an elliptical cone beam. It is the preferred low-cost solution for obstacle avoidance of AGVs with low horizontal surfaces and fixed path navigation.

The main features are as follows

1° high angular resolution, accurate identification of obstacle locations

Custom obstacle avoidance mode and secondary development obstacle avoidance mode are optional

20Hz initial detection frequency, faster obstacle avoidance response

Small size, with built-in mounting holes, can be directly embedded in the machine

There are no rotating parts inside, the structure is reliable and stable

No echo interference or ground inequality false alarms

Figure 4 Schematic diagram of 3D array directional obstacle avoidance laser radar

Figure 5 Schematic diagram of custom obstacle avoidance mode (rectangular area in the figure)

The custom obstacle avoidance mode is a mode developed specifically for AGV obstacle avoidance applications. As shown in Figure 5, in this mode, the 3D array laser radar will screen out the most critical obstacle avoidance target for the AGV, that is, target A in the warning area, and provide the distance information of this target to the AGV. The obstacle avoidance area can be customized according to customer needs.

Figure 6 Secondary development of obstacle avoidance mode (comparison between the actual scene image (top) and the depth image (middle) and point cloud image (bottom) taken by the radar)

The secondary development obstacle avoidance mode provides the AGV with a point cloud map of obstacles to "depict" the general outline of the obstacles. Under default conditions, the 3D array laser radar will output a depth map and the corresponding signal strength data for each pixel. The depth value of each pixel in the depth map represents the projection distance from the detection point corresponding to the pixel to the front surface of the radar. As shown in Figure 6, the actual scene image taken by an ordinary grayscale camera (top), the depth map taken by the 3D array laser radar at the same position (middle), and the point cloud map drawn based on the depth data (bottom, observed from a position slightly higher than the 3D array laser radar) are shown respectively.

02

LiDAR application cases in AGV and warehousing

Warehouse AGV obstacle avoidance

A 3D area array LiDAR is installed directly in front of the warehouse AGV to monitor obstacles in front in real time, control the AGV to slow down or brake, and assist in its quick storage and retrieval, intelligent transportation and other functions.

Solution advantages: Supports multi-machine collaboration, can resist interference from ambient light in the warehouse, has a structure that is not easily damaged, and is reliable and stable.

Warehouse forklift obstacle avoidance

The laser radar is installed in front of the forklift fork to identify the distance of the obstacle in front and feed it back to the forklift. After the forklift system processes it, it performs anti-collision or assists in locating the pallet.

Solution advantages: We have been deeply involved in the industrial control industry for a long time, and can be customized for customers with high matching degree. We are cost-effective in terms of low-cost obstacle avoidance.

Obstacle avoidance in high-rise warehouse

Two laser radars are installed on the front and rear of the smart shuttle respectively to monitor the front and rear obstacles and distance information in real time, and control the car to slow down or stop suddenly, thereby realizing functions such as storage and retrieval, transportation, and obstacle avoidance.

Solution advantages: can be directly installed on the car, easy to install, quick to debug, cost-effective, accurate obstacle avoidance.

兰博

I have a robot vacuum cleaner at home that can avoid obstacles.

But in fact, it is very stupid. When encountering complex terrain, it often cannot get out. It seems that there is still a long way to go before this thing can be developed and matured.