[Shenzhen University Electric Competition] Simple motion capture gloves

This project won the first prize in the 2020 Shenzhen University Electronic Design Competition

Title: Simple motion capture glove design (microcontroller embedded control)

• Design and make a simple motion capture glove to capture the motion of human finger joints.

1. Basic parts:

• Design and make a motion capture glove (for both left and right hands). The hardware device on the glove can be fixed on the glove body. It is lightweight and can work for a long time.

• Motion capture gloves must have a calibration function, that is, they must be placed on a tabletop before wearing to correct sensor errors. The motion capture gloves can detect the bending of at least two joints of each finger in real time and can transmit the data to the computer through a wireless module.

• Design a computer-side computer program for displaying three-dimensional movements of human hands, which can communicate wirelessly with motion capture gloves, and convert the data sent by the gloves into synchronized human hand movement states in the program.

2. Play part:

• In addition to detecting the bending of finger joints, the motion capture gloves can also detect the movement of the human hand and forearm in space, and display it in the host computer program.

3. Special instructions

• Sensors on the gloves can use micro-inertial measurement units, bend sensors, etc. to detect joints.

2020 Shenzhen University Electronic Design Competition Final Report

1. demand analysis

• The topic requires the production of a simple motion capture glove, which requires the recognition of at least two joints for each finger, and the corresponding 3D model can be displayed through the host computer.
• The structure of human fingers is relatively complex and consists of multiple joints. Therefore, multiple sensors need to be placed at key locations on the hand to collect movement information of each finger and joint.
• We can think of the handle as a model with multiple vectors connected end to end. Take a finger as an example as shown below.

• Vector A and vector B are two vectors of fixed length. They are connected end to end and represent two finger joints. The point where AB is connected is the finger joint.
• If you need to know the state of two vectors in space, there are two ideas:
  • Know the angle α and angle γ;
  • Know the angle α and β.
• Starting from one of the above two ideas, the position of the two vectors in space can be determined through an algorithm. To construct a hand model in the same way, all that needs to be done is to add vectors and connection relationships, and extend it to three dimensions.

2. Plan selection and demonstration

Solution 1: Based on stm32 as the main control, using multiple bending sensors.

• Multiple bending sensors are used to collect the bending degree between finger knuckles and convert it into the relative angle between the knuckles (i.e., the β angle in the above figure).
• Advantages: Program and circuit design are convenient.
• Disadvantages: The bending sensor is expensive and is too expensive to make into a glove. A bending sensor can only identify the rotation of one rotation axis. Some joints of the hand can rotate in two directions. This part of the rotation needs to be identified by adding a bending sensor.

Option 2: Based on stm32 as the main control, using multiple MPU6050 sensors.

• Multiple MPU6050s are used to collect the angle between each finger knuckle and the world coordinate, and the motion model of the hand is restored through conversion and data processing.
• Advantages: Each MPU6050 can identify angle changes on three rotation axes, so it can respond to hand movements more accurately, and the cost is lower than using bending sensors.
• Disadvantages: There are many circuit boards, there are some difficulties in wiring, and the program design is slightly more complicated than the other two solutions.

Option 3: Based on stm32 as the main control, multiple potentiometers are used to identify the finger angle:

• Multiple potentiometers are linked to the finger joints, and the partial voltage value is obtained through the ADC and then converted into an angle to restore the motion model of the hand.
• Advantages: low price, simple circuit, simple programming
• Disadvantages: The gloves have a complex design and structure and are easily damaged.

After comprehensive consideration, option 2 is adopted as the sensor solution. Other functional implementation plans are as follows:

3． System architecture and function description

4. Theoretical analysis and calculation

• Import legend:

5. Circuit diagram and PCB

• Power circuit:

• Main control chip circuit:

• Electricity measurement circuit:

• LED circuit: This work has a power indicator light, a power indicator light (lights up when the battery needs charging), and two spare indicator lights.

• Bluetooth circuit: The gloves need to exchange data with the computer. This work chooses the Bluetooth module HC08 to transmit data. The following is part of the Bluetooth circuit.

• Six-axis circuit:

• The picture below shows the six-axis circuit of the wrist and upper joints.

• The picture below shows the six-axis circuit of the lower joint.

• The picture below shows the six axes on the main control.

• The picture below shows the six-axis interface.

6. Debugging and test results

• The first step: first weld the microcontroller main control board and test its hardware circuit to test whether the voltage of each module is normal, whether the boost and buck chips, power management chips, etc. are working properly. The test results are normal.
• Step 2: Program a simple routine for the microcontroller and test whether the microcontroller can work normally. The test results are normal.
• Step 3: Use the microcontroller to read the MPU6050 on the main control board and test whether it is working normally. The test result is normal.
• Step 4: Weld the MPU6050 small modules on each finger, test the hardware for short circuit and open circuit, and test whether the microcontroller can be used to communicate with it normally. All MPU6050 tests are normal.
• Step 5: Test and configure the Bluetooth communication chip. The test is normal.

• The first step: read the data of all MPU6050 modules, package it and send it to the PC host computer through Bluetooth.
• Step 2: Use Unity3D to write a program on the PC host computer, call the serial port API provided by the .NET framework to read and analyze the data, and map the data to the 3D model.
• Step 3: Optimize the program.

7． Demo video