Control Segbot with TV Remote

This project is the final project of ME461 of UIUC.

In this project, I used an IR receiver to receive the signal from a TV remote controller to control the motion of the Segbot. I'm also considering changing the control gain with the remote controller in the future.

Motivation

Segbot is a self-balancing robot designed by Prof. Dan Block(d-block@illinois.edu) controlled by TMS320F28379D from Texas Instrument. We are given a set of control gains to make the Segbot balance on itself. However, during the process of finding out the optimum values for a more responsive result, I found that the behavior of Segbot will be affected by the USB cable connected to it (which is heavy for the Segbot, but necessary to change the values and see if that works better). I then found that using a TV controller and an IR sensor may help achieve that.

Build Process

I soldered an IR sensor and an ultrasonic sensor on a prototype board and put it at the top of my Segbot. The photo of Segbot and connection detail is shown below.

Working demo

There are 6 commands for the Segbot in this project.

• Move forward
• Move backward
• Turn left at a rate (increase the left turn rate)
• Turn right at a rate (increase the right turn rate)
• Turn a step left (jump left)
• Turn a step right (jump right)

Below is a figure of the corresponding button

Here is a video of controlling the Segbot. The "strange" noise in the video is from the ultrasonic sensor.

Reading Sensor Data

To use the TV remote, we need to figure out how to decode the remote signal using our F28379D processor. In this project, I first figured out the output pattern from the IR receiver when receiving a signal from the TV remote (shown below). After figuring out that, I sampled the signal with a proper rate, which is 0.1 ms in my project. Then I used a state machine to decode the output. A more detailed version of tutorial could be found here.

For the ultrasonic sensor, we need to trigger it by a pulse for at least 10 microseconds. In this project, I choose to use the trigger time of 10ms and an interval of 5 ms.

Then we need to use the eCAP peripheral to capture the echo result. Since we know the sound speed, we can get the distance of the object by 0.5*echo_time*340m/s.

The below code is the trigger function for the HC-SR04 ultrasonic sensor. In this function, I make GPIO0 to stay high for 10ms and stay low for 5ms every time the function is called.

//function to trigger ultrasonic sensor
//each trigger takes 15ms
void hc_sr04_trigger(void) {
    if (trigger_count<5) {
        GpioDataRegs.GPACLEAR.bit.GPIO0 = 1;//stay low for 5 ms
    } else {
        GpioDataRegs.GPASET.bit.GPIO0 = 1;//stay high for 10 ms
    }
    trigger_count = (trigger_count+1)%14;
}

In this project, whenever the Segbot finds an obstacle within 30 cm, it will automatically back up and turn left for 1.5 seconds. After that, it will stop. This is achieved by a simple state machine shown in the following code in the software interrupt.

//check if there's an obstacle within 30cm
//If so, back up and turn, check again after 1.5s
    if(trigger_state == 0){
        if(echo_distance < 30) {
            trigger_state = 1;
            trigger_num = 0;
        }
    } else if (trigger_state == 1) {
        FwdBackOffset = 0.3;
        turnrate = -0.3;
        if (trigger_num >= 1500) {//trigger_num increase 1 every 1ms
            trigger_state = 0;
            FwdBackOffset = -0.15; //after 1.5s stop backing up and go forward
            turnrate = 0;
        }
    }

Credits

Thanks Yorihisa Yamamoto for the LEGO Segbot model; thanks Professor Dan Block, TA Chad Li, and Hang Cui for this wonderful semester!