Project
# | Title | Team Members | TA | Documents | Sponsor |
---|---|---|---|---|---|
12 | Shoe Sorting Robot |
Jinghan Guo Mingxi Zou Quanhua Huang |
Zhen Qin | design_document1.pdf final_paper1.pdf other2.pdf other3.pdf other1.pdf presentation1.pdf proposal1.pdf |
|
Have you ever get tripped over a shoe in the doorway? Disorganized shoes can be a mess but sometimes people just don’t get time to organize after they take shoes off. Therefore, Our group aims to create a robot that helps people sort shoes and save their time. So far there is no such robot on the market, therefore we believe this project will be novel and it will become useful to keep house clean. In our project, we will connect a CMOS camera module, a robot arm, a load cell amplifier ADC weight sensor and a power control to our own microcontroller. Our robot will be built based on the scenario that shoes are scattered on a 60*60cm entrance mat with a rectangular shoe organizer next to it. The camera will be held above at certain height to capture the image of whole mat. The 6 degree of freedom robot arm will be attached to a mobile car that is rested next to the shelf if there is no shoes on the mat. After the camera detects more than two individual objects (one pair of shoe), the car will start to move and the robot will organize the shoes. Our robot will distinguish each shoe by color and weight, and place each pair on the organizer from the leftmost corner. From camera we can know the size (how many pixels) of the object, if it lies in certain range (from size 5 shoe to size 9 for example), we assign it as a "shoe". Then we use the RGB value to assign the color of the shoe. Among all the white shoes for example, we can decide if two shoes are a pair depend on their size and weight. As a result, those shoes will be arranged in pairs and pairs in similar color will be placed next to each other. Considering about situations that objects other than shoes may appear on the map, we will implement color filtering and weight filtering. If the size of the objects appear to be too large or too small, we just ignore them. If the objects have similar size but the robot cannot find another paired object, the robot will still grab the umbrella and put it on the shelf in the end in order to keep the doorway clean. The shoes sorting algorithm is described as follows: 1.Camera locating on top of the mat scans all the objects(shoes) and then marks down each object. 2. Calculate the coordinates of the center of each object with respect to the world frame coordinates, where we build the world frame using the robot’s center. 3.Assign shoes with same color a specific number according to RGB value. 4.Robot arm randomly picks a shoe with a certain color, for example white, and places it in the position #1 of the shelf. At the same time, the weight of that shoe will be detected by the weight sensor connected to the gripper and saved. 5.Robot arm will look for the other white shoes and pick them up one by one. If the white shoe has the same weight as a shoe that is already on the shelf, then these two shoes are considered to be a pair and will be placed together. If the white shoe has different weight, then we will place this shoe in the subsequent position on the shelf. The process continues until all the white shoes are ordered on the shelf. After the white shoes have been sorted, the robot arm will proceed to sort the remaining shoes according to their colors. We think that even if no shoes are exactly the same weight, two shoes in a pair will be very similar. Our robot arm is able to lift something at least 500 grams and we decide to assign an offset of +/-5%. If two shoes has approximately the same weight then we assign them as a pair. The end effector of the robot arm is indeed a consider of our project, we are planning to order few different types of end effector to see which one grabs the shoe the best. Our group have measured the weight of 5 different pairs of shoes, including sneakers and boots, and the heaviest boot is less than 400g, and most of the robot arm on market is able to lift about 500g, therefore weight may not be our biggest concern, but at which location of the shoe should the end effector grab the shoe so the shoe won't fall on the robot's way back to the shelf. We believe we will need a lot of trials to come up with a good solution. |