Plant Nursery Bot (2019)

#Robotics  #Solidworks  #Arduino  #C++  #3D-Printing  #LaserCutting

redbox.png

INTRODUCTION

1. Detection    2. Collection    3. Storage    4. Transportation 

Existing plant management and harvesting approaches, such as visual scouting, ground-based proximal sensing, and manual harvesting are labor-intensive, time-consuming, and do not scale quickly to large areas. Therefore, to facilitate more efficient farming methods, innovative technologies must be adopted to achieve optimal plant management as well as giving rise to the rate of food production. The Robotic Design Competition held annually by the American Society of Agricultural and Biological Engineers (ASABE) intends to identify such technologies while inspiring young professionals to develop skills in robotic systems by simulating a fully autonomous solution to a common agricultural process.  

 

For the 2019 competition, participants were challenged to design a robot capable of conducting autonomous inventory management in a plant nursery simulated by a 4 ft. x 16 ft. board. During the competition, two teams simultaneously competed with one another to correctly identify and arrange the most nursery plants in the shortest possible time within a five-minute time frame. The figure below shows the layout of the field. There were a total of 21 simulated red and green pot-plants located on the board. The robots needed to autonomously identify, collect, and transport these pots to their respective storage area on the other side of the board. The pots were required to remain upright throughout the whole process. One plant, the “Golden Pot” was placed in the center of the board and served as the bonus points for the team that collected it first. The robots were restricted to be within a 12 in. x 12 in. x 12 in. dimensions at the beginning of the competition. With the exception of the Golden Pot, all other plants were randomly placed on the board as shown below.

 

 

The main task in this project was to design and develop an autonomous system capable of finding the plant, distinguish if it is red or green, and place it in the correct storage location. This project was accomplished in the spring of 2019 with the assistance of the HuskerBot Robotics team, which was lead by me and included the following members: Cheetown Liew, Abbas Atefi, Aaron Haake, Herve Mwunguzi, Jay Min Lim, Pranav Palli, and Jun Hong Vince Chong. The project was guided by Dr. Santosh Pitla, associate professor at the Department of Biological System Engineering at the University of Nebraska, Lincoln. The first step towards building this system was to brainstorm different concepts and construct the initial mechanical design. With the restrictions imposed for this challenge, various design options were proposed, some of which included a conveyor belt system and circular multi-arm robot. The design methodology adopted by the team can be summarized in the following areas:

 

 

board.png
ASABE poster2box.png
In order to achieve the differentiation between the two categories of red and green plants, the robotic system relied on an RGB (Adafruit TCS34725) sensor. This sensor is capable of quickly detecting colored objects within 1.5 inches of its proximity, making it ideal for this project. 
To pick up and deposit the pots, a custom-designed robotic arm consisting of a pair of parallel four-bar linkage was developed. This robotic arm is controlled by two hobbyist servos-one for rotating the arm horizontally and the other to move it vertically. A 360-degree rotation servo (Parallax Feedback 360° High-Speed Servo) was implemented for ease of movement during the placement and removal of the collected pots in the storage unit. The direct collection and deposition process is handled by a High-Torque 2BB Metal Gear Servo-Hitec HS-645MG. This allows for unconstrained movement and smooth operation. These servos are directly controlled by a Lynxmotion servo controller (Lynxmotion SSC-32). Commands are sent to this board via serial communication from the Arduino.
The mobile platform is built on a 4WD 60mm Mecanum Wheel Chassis. This specification allows for the robot to move in all directions without having to turn to the respective angle, thus reducing the time required for steering. The platform is controlled by high-reduction geared DC motors to enhance the smoothness of motion. Adafruit HC-SR04 Ultrasonic sensor, one at each side of the robot were implemented to help navigate the chassis around the board. Moreover, a line following sensors was also used to track the pots along the line where they were set up.   
To store the pots, two circular units, each capable of holding up to five pots were designed. The holders were engraved to a small depth to keep the pots upright at all times. Each unit is driven by a 360-degree rotating servo, which enables a continuous rotation. Initially, the plates are folded, and upon initiation of the system, they unfold simultaneously.
1. Detection
2. Collection
3. Storage
4. Transportation

The robotic system at hand utilizes ultrasonic sensors to navigate its way through a 16 ft. x 4 ft. simulated environment, which represents a plant nursery. As soon as the robot starts running, it first positions itself along the row where the pots are placed, unfolds the storage units, and steers forwards until the ultrasonic sensor detects the first pots at the 10-inch distance. The mechanism will then pause for the custom-designed robotic arm to reach the pot. The gripper of the arm is initially open to allow the arm to position itself around it. The gripper then closes, allowing the RGB sensor to get closer to the plant and detect its color. Once the correct colors have been identified, the arm transfers the pot plants to either the right plate, which holds the green pots or the left plate, which is to store the red pots. Finally, with the assistance of the mecanum wheels, the system smoothly rotates and steers towards the storage area to dispatch the pot plants in the two marked zones. The average time recorded for the successful collection or dispatch of a pot was measured to be 12 seconds. This is measured from the period the system is positioned correctly in front of the pot all the way to the correct placement of the object in its storage unit or the reverse process. The total time to collect all the pots successfully, starting from the corner of the board until the robot reaches the center of the simulated area was averaged at 137 seconds. This time also accounts for to initial positioning of the system as well as allowing the unfolding process of the storage plates. The average time taken for the robot to rotates from the center, travel to the other end and complete the dispatch process was 132 seconds. Therefore, the total time taken for a truly successful run was found to be 269 seconds (4 minutes and 48 seconds), which was within the time frame of the competition.

redbox.png

INTEGRATION