Smart Object Identifying Robot
In the past, the work done on object detecting robots they are not efficient. Those robots are not fully functional. The main problem of object detection is that the robot will start detecting and recognize the object and give that information to the user. We are adding intelligence to it that inclu
2025-06-28 16:29:22 - Adil Khan
Smart Object Identifying Robot
Project Area of Specialization Artificial IntelligenceProject SummaryIn the past, the work done on object detecting robots they are not efficient. Those robots are not fully functional. The main problem of object detection is that the robot will start detecting and recognize the object and give that information to the user. We are adding intelligence to it that includes object detection as well as it will be navigational. So far, for this problem, there is no productive and effective solution.
The research made in this field, methods developed were not suitable for real-time application.
As we dive into the details of robotics and object detection, we see that these are two different fields but when it comes to the implementation, they form a huge connection. A robot that detects objects and moves autonomously involves artificial intelligence-related algorithms. Autonomous robots are able to gather information from their surroundings and are they are able to work for extended periods without human interference. Furthermore, these robots can operate themselves throughout operation without human assistance.
Project Objectives1. The motive of this project is to build a robot that performs object detection and recognition in real-time video capturing.
2. The robot will be able to locate the objects and navigate on a defined path. It will act independently on the defined area on which we have trained our robot and show us a list of detected objects.
3. Furthermore, the robot will be able to move in all possible directions. The movements of the robot are controlled by NVIDIA Jetson Nano.
Here we will discuss the detailed methodology of the robot. The detailed explanation of the work doneu p till now. Moreover, the implementation techniques of object detection are well explained and the construction of robot up till now is also mentioned.
The approach we proposed is convolutional neural network (CNN/Conv-Net) that is a class of deep neuraln etworks, most commonly used to analyze visual imagery. Now when we think of a neural network we think about matrix multiplications but that is not the case with Conv-Net. It uses a special technique called Convolution. In mathematics convolution is a mathematical operation which performed on twof unctions that produces a third function that shows how the shape of one is modified by the other. There are three important steps which are discussed below :
1. Convolutional operation
2. Pooling
3. Flattening
3.2.1 YOLO Algorithm
Yolo stands for you only look once mean yolo algorithm look an image just for once. To understand
this algorithm, it is necessary to establish what is actually being predicted. Ultimately, we aim to predict
a class of an object and bounding nox specifying object location. Each bounding box can be described
using four descriptors:
1. Center of bounding box (bx,by)
2. Width (bw)
3. Height (bh)
4. Value of c which is corresponding to a class of an object such as cars, building cats, human traffic
light etc. In addition, we need to predict pc value, which is the probability that there is an object
in bounding box or not.
Y= (pc,bx,by,bw,bh,c)
3.2.2 Working of YOLO
Yolo algorithm predict objects and bounding boxes. First of all, it divides the given image into SxS size
(usually 19x19) or cells and every cell can detect B bounding boxes whose size is smaller than the cell
with a class probability. YOLO algorithm detects bounding boxes with class probability (x, y, w, h, c,
confidence).
3.2.3 Intersection over Union
YOLO algorithm have some ground truth of the objects, it takes intersection over union of cell and ground
truth to get the exact position of object. IOU is to check that how much ground truth is overlapped with
predicted one. Then we find out the confidence of that cell by multiplying the probability of object
wit IOU. If confidence value is 0 it means there is no object in cell so no need to proceed that cell, if
confidence is 1 it means there is an object in cell so we proceed further steps on that cell.
3.2.4 Training the model
The training data se have images with labels. In yolo algorithm labels are bounding boxes (x, y, w, h,
c, confidence) values so its bounding box coordinates and the class labels. Why is it so fast? Because
there is no pre-processing of the input images and there is just a single forward propagation.
The benefits of this project is that it can be used where human cannot reach or assist. Furthermore, we have added autonomous part in it in which we set the starting point and destination point and robot will reach to it's destination by itself. That is the major benefit of this project, we have integrated object detection as well as autonomously working in one robot.
Technical Details of Final DeliverableFinal deliverables will be a robot containing all the components required for making it fully functional. The structure of robot is a rocker bogie. The object detection is done on NVIDIA jetson Nano developer kit. The camera will be mounted on the robot. While robot moving to it's destination it will detect all the objects that will come in its way and send the names of detected objects to the user.
Final Deliverable of the Project HW/SW integrated systemCore Industry ITOther IndustriesCore Technology Artificial Intelligence(AI)Other TechnologiesSustainable Development Goals Industry, Innovation and InfrastructureRequired Resources| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 161397 | |||
| NIVIDIA Jetson Nano Developer Kit | Equipment | 1 | 27899 | 27899 |
| Dc motors | Equipment | 6 | 1000 | 6000 |
| LCD for robot | Equipment | 1 | 12000 | 12000 |
| IBT2 Motor Driver | Equipment | 2 | 1450 | 2900 |
| Arduino Uno | Equipment | 1 | 3000 | 3000 |
| Robot Chassis | Equipment | 1 | 2000 | 2000 |
| NIVIDIA Jetson Nano Developer Kit | Equipment | 1 | 27899 | 27899 |
| Dc motors | Equipment | 6 | 1000 | 6000 |
| LCD for robot | Equipment | 1 | 12000 | 12000 |
| IBT2 Motor Driver | Equipment | 2 | 1450 | 2900 |
| Arduino Uno | Equipment | 1 | 3000 | 3000 |
| Robot Chassis | Equipment | 1 | 2000 | 2000 |
| NIVIDIA Jetson Nano Developer Kit | Equipment | 1 | 27899 | 27899 |
| Dc motors | Equipment | 6 | 1000 | 6000 |
| LCD for robot | Equipment | 1 | 12000 | 12000 |
| IBT2 Motor Driver | Equipment | 2 | 1450 | 2900 |
| Arduino Uno | Equipment | 1 | 3000 | 3000 |
| Robot Chassis | Equipment | 1 | 2000 | 2000 |