Project

# Title Team Members TA Documents Sponsor
23 FPGA-based object tracking, obstacle avoidance, and voice-activated trolley
 Haomin Wang
Jiarun Hu
Yang Zhou
Yihang He
 Tielong Cai design_document3.pdf
final_paper2.pdf
proposal1.pdf
 Said Mikki
# Members:

- Yang Zhou [yangz15]

- Haomin Wang [haominw3]

- Yihang He [yihangh2]

- Jiarun Hu [jiarunh2]

# Problem:
Nowadays the development of electric vehicles today has become a trend. At the same time, more and more new energy vehicle startups like to equip their cars with intelligent systems. However, existing SOCs are always based on non-real-time operating systems and need to meet the real-time property and safety of the in-vehicle system. Common systems which are based on CPU + GPU tend to have high energy consumption, which will ha a negative impact on the endurance of the vehicle. Therefore, designing a system with low energy consumption and high real-time performance is necessary.

# Solution Overview

In order to achieve low energy consumption and high real-time performance, our solution is to design a specific system to control our trolley based on FPGA, which combines four subsystems. The first subsystem processes real-time data from the other subsystems to control the trolley. The second subsystem is designed to detect the target object and send a tracking signal to the movement control subsystem. The third subsystem is to detect obstacles in the path of the trolley and send an avoidance signal to the first one. The last subsystem is to recognize natural language instructions from the operator and sends the corresponding signal to the movement control subsystem. By taking these four aspects into account, we will create our object tracking, obstacle avoidance, and voice-activated trolley.

# Solution components:

1. **Trolley movement control subsystem:** The movement control subsystem will process real-time data from the other subsystems and produce the signal to control the movement of the trolley. Control signals will be passed through the FPGA port to the PCB board, which is connected to electric motors. The PCB board can generate current to control the speed of electric motors depending on the control signal so that our trolley can move as designed. 

2. **Object tracking subsystem:** The object tracking subsystem will use a camera to catch the image in front of the trolley. FPGA will receive the image and process it to identify the location of the color block and generate suitable control signals based on the location of the color block so that the trolley can move toward the color block.

3. **Obstacle avoidance subsystem:** We will use ultrasonic sensors to detect obstacles in the path of the trolley. The FPGA will be used to process the signals from the sensors and control the movement of the trolley. The microcontroller should be programmed with algorithms for obstacle detection and avoidance.

4. **Voice-activated subsystem:** Our design target is that the trolley can recognize specific natural language instructions and act accordingly. Thus, we will design a voice-activated system and combine it with the control system of the trolley. In order to reduce the latency as well as achieve high recognition accuracy, we will build a CNN network on FPGA instead of LSTM or DSP procedure to do this task. And this voice-activated system will give the corresponding signal to the control part.

# CRITERION FOR SUCCESS:

1. The trolley should be able to move at a reasonable speed so that it can avoid obstacles and respond to voice commands in a timely manner. The movement control subsystem will also be able to process conflicting instructions and produce the correct signal to control the movement. The subsystem needs to be secure and reliable. 
2. The trolley should be able to use a camera to detect a color block and move toward the color block. This can be measured by testing if the trolley can follow the movement of the color block closely.
3. The trolley should be able to detect obstacles accurately and reliably using its sensors and cameras. This can be measured by testing the trolley's ability to detect and avoid obstacles of different sizes and shapes. 
4. The trolley should be able to recognize and respond to specific voice commands accurately and reliably. This can be measured by testing the trolley's ability to understand a range of voice commands and respond accordingly.

# DISTRIBUTION OF WORK:

## Yang Zhou, Electrical Engineering:
Design and implement the trolley movement subsystem. Implement and test the way control subsystems interact with other subsystems.

## Haomin Wang, Computer Engineering:
Design and implement the object tracking subsystem. Test the trolley's ability to detect and follow the color block.

## Yihang He, Computer Engineering:
Design and implement the obstacle avoidance subsystem. Test the trolley's ability to detect and avoid obstacles of different sizes and shapes.

## Jiarun Hu, Electrical Engineering:

Design and implement the Voice-activated subsystem. Test the trolley's ability to recognize natural language instruction and control the movement of the trolley.

A Micro-Tribotester to Characterize the Wear Phenomenon

Shuren Li, Boyang Shen, Sirui Wang, Ze Wang

A Micro-Tribotester to Characterize the Wear Phenomenon

Featured Project

**Problem**

Many research efforts have been made to understand the complex wear mechanisms used to reduce wear in sliding systems and thus reduce industrial losses. To characterize the wear process, coefficient of friction needs to be measured “not only after completion of the wear test but also during the wear test to understand the transitional wear behavior that led to the final state”.(Penkov) In order to improve the effectiveness and efficiency of these research methods, it is necessary to improve the instrument used to characterize the wear phenomenon to better measure the friction coefficient of the material. Although the instrument can be applied on all solid samples, we will use silicon wafer coated with SiO2 as our specimen targeted object.

**Solution Overview**

The objective of the experiment is to evaluate the wear phenomenon of the sample during the sliding test so as to obtain the wear information of the material. We will design planar positioning and force sensing system to get the move and force information of our objects. To collect the data of vertical load and horizontal friction, 2 force sensors are mounted on linear rails to minimize the radial force and ensure that only the axial forces are collected. Then, the coefficient of friction can be calculated by equation:

![](https://courses.grainger.illinois.edu/ece445zjui/pace/getfile/18615)

And to determine the relationship between the coefficient of friction and the state of wear, we use a microscope to monitor the state of wear at a given location in the wear track and evaluate the wear process during each sliding cycle. In this way, we can investigate the wear transition processes with respect to the sliding distance then transport our data to a computer. Finally, we will design our data processing method in the computer to successfully obtain an acceptable result in the margin error.

**Solution Components**

1. Motion Platform: This subsystem includes a linear actuator that moves the sample in reciprocating motion along X-axis, a stationary counter surface that applies constant vertical load onto the sample, and another actuator that compresses the spring and provides a vertical load to the counter sample.

2. Specimen and Counter surface: We will test the wear and friction between the specimen and the counter surface during the sliding test. A 10 × 10 mm^2 silicon (Si) wafer coated with 50 nm thick SiO2 will be used as the specimen and a stainless-steel ball with a diameter of 1 mm was used as the counter surface.

3. Sensors: This subsystem includes two force sensors that measure the vertical load and horizontal friction. The Load Sensor should assemble along with the Z-axis actuator. To measure the friction without the effect of load, we assemble the Load Sensor and Friction Sensor sensor on the Linear Rails, as the photo attached shows. Since the sensors are strain gauges and only outputs, small changes in resistance, amplifiers, and ADC are needed to collect the signal and send converted data to the computer.

4. Data Processing: This subsystem includes acquiring raw data of load and friction on the computer, applying necessary filters to reduce noise and improve accuracy, and plotting the result that reflects the relationship between the sliding cycles and coefficient of friction for our sample.

![](https://courses.grainger.illinois.edu/ece445zjui/pace/getfile/18611)

**Criterion for Success**

1. Motion platform can perform precise reciprocation. The control system can effectively control the number and speed of reciprocating motion.

2. The acquisition unit can collect data effectively and can transfer the data that can be processed to the computer.

3. On a computer, the raw data can be processed into a readable graph based on algorithms set up. By analyzing the graph, the relationship between the data and the expected results can be correctly obtained.

**References**

Penkov OV, Khadem M, Nieto A, Kim T-H, Kim D-E. Design and Construction of a Micro-Tribotester for Precise In-Situ Wear Measurements. Micromachines. 2017; 8(4):103. https://doi.org/10.3390/mi8040103