R.O.B. (EECS 373 W'08)

Team Members:

Vishnu Desaraju (rajeswar)
Bobby Li (bobbyli)
Honie Lui (homedogg)
Zane Salim (zsalim)

___________________________________

Introduction:

Our project idea was to design a remote controlled car that can fire darts from a movable launcher and also sense the distance to any object in front of it.

The car is controlled using a mobile MPC-555 processor and is driven using a wireless Playstation 2 controller. The dart launcher carries two darts and can be rotated in two directions (pitch and yaw). It also has an infrared distance sensor mounted on its front to measure distance to its target.

Sensor and current payload data is sent back to the base station (the MPC-823 processor and Spartan-3 FPGA) using a wireless serial link. The base then displays the distance and the dart count on a character display.

___________________________________

***Member* *Task Distribution:***

Vishnu Desaraju
H-Bridge Interfacing
Servo Calibration
MPC-555 Data Pin Connections
PS2 Controller Interfacing (Protocol & Interface)
Integration (K’nex Expert, Dart Launcher, Circuits)

Bobby Li
Character Display
Wireless Transmission
PS2 Controller (Protocol & Interface)
Distance Sensor (MPC-555 ADC)
Integration (Wireless Integration, Spoiler)

Honie Lui
Distance Sensor (MPC-555 ADC)
Wireless Transmission
Servo Calibration
H-Bridge Interfacing
PS2 Controller (Protocol)
Integration (Car Chassis, Dart Launcher, Circuits, Wireless Integration)

Zane Salim
Character Display
Wireless Transmission
Servo Calibration
PS2 Controller (Protocol)
Integration (Circuits, Integration Code, Dart Launcher, Wireless Integration)

High Level Design:

Block Diagram

High Level Diagram

Infrared Distance Sensor
An infrared distance sensor was mounted to the front of the car to determine the distance between the car (specifically, the dart launcher) and the object that the sensor was pointed towards. This information was processed by the MPC-555 and then sent wirelessly to the MPC-823.

RF Wireless Link
The transmitter was connected to the MPC-555 and mounted on top of the car. The MPC-555 would relay information on distance and ammo via a serial interface to this transmitter. The receiver was connected via a serial interface to the MPC-823, which would receive this data wirelessly for processing and writing out to the character display.

Servos
The servos that were used provided us with a +/- 90 degree displacement from center, and enabled us to use one servo to steer the car and three servos to control the dart launcher. For the dart launcher, one servo was used to launch a dart, another was used to rotate the launchers left or right, and the last servo was used to adjust the angle / tilt of the dart launchers.

Character Display
A character display was used to display the data from the distance sensor and the ammo remaining. The data was received through a serial bus with the wireless link. The MPC-823 would then process the information and transmit it to the character display through the FPGA.

Wireless Playstation 2 Controller:
A chameleon wireless Playstation 2 controller was used to control the car and the dart launcher. The controller provided buttons for accelerating, reversing, aiming the dart launcher (rotation and tilt), and firing the left and/or right dart.

Car Motor:
This motor was made to run on 9.6 volts. We were able to accelerate the motor in two directions (forwards and backwards) by using an H-Bridge chip and driving out certain signals from the MPC-555.

Hardware Design:

Infrared Distance Sensor
The distance sensor used had a range of 8 inches to 60 inches. The output from the distance sensor was connected to one of the MPC-555 queued analog to digital converter pins. We were able to access the register in C that contained this voltage information, and then characterize the sensor by using several meter sticks and some reflective tape. From this, we were able to make a general mapping of what voltages corresponded to what distances.

RF Wireless Link
On the wireless transmission side, the MPC-555 used a serial interface to drive data to the transmitter with a baud rate of 1200. The MAX232 chip was used to convert the serial voltages (-10 V to 10 V) coming from the MPC-555 serial port to a 0-5 volt signal that the transmitter was able to process. On the wireless receiving side, the signal expected was 0-5 V. We saw that the signal became slightly weaker here (not quite dropping to 0 volts), and used this weak signal to drive a bipolar junction transistor / inverter pair to get a proper 0-5 V data signal. This data signal would then be converted back to serial voltages using the MAX232 chip and finally sent to the MPC-823 via serial.

Servos
We used a total of four servos in our final design: one for steering, two for dart aiming, and the last to fire the darts. The servos had a range of roughly +/- 90 degrees. Each was controlled using a dedicated PWM output pin on the MPC-555. We also had to use a prescaler to slow down the effective system clock for the PWM to get pulse widths in the range of ~0.7ms to ~1.7ms.

H-Bridge and Motor
The DC motor was included in the car chassis, but in order to drive it forward and in reverse, we had to drive the motor through an H-Bridge circuit. We used a Quadruple Half-H chip and connected our PWM to its enable line. We also connected two input channels on the chip to output pins on the MPC-555. Driving one pin high and the other low would then spin the motor in one direction, while reversing those values would make the motor spin the opposite direction.

Character Display
The character display was used to display the distance obtained from the infrared sensor on the car and the amount of ammo remaining from the dart launcher. The FPGA was programmed with the logic needed to interface with the display. The display had 3 control pins, 8 data pins, and 3 power pins. The write enable for the outputs on the expansion board used for the control pins were only high when we wanted to actually write to the display. At first, the write enable was always high but the display did not give any type of response. We think this is because when we are not writing to the display, the display is driving those pins with values such as the busy flag.

The 3 control pins were RS, R/W, and E. The write enable for these 3 was wired to Vdd since we would always be driving these pins. RS and R/W were latched by the display on the rising edge of E and the data pins were latched on the falling edge of E. Therefore, the setup and hold times of RS, R/W and data were important, as well as the pulse width of E. We took the times given to us by the spec and doubled them to absolutely make sure we fulfilled the requirements. Our character display also used the busy flag. To do this, we had to change the write enable on the expansion board to be controlled by DIR instead of the switches. A wire was placed between one of the expansion board outputs and the DIR input to switch the write enable on and off.

PS2 Controller
We used a wireless PS2 Controller provided by Matt. It was used to control the movement of the car, the direction of the dart launcher, and the firing of the dart launcher. The controller has 9 pins, 2 of which are used for power and ground. The other 7 pins are DATA, CMD, SEL, CLK, ACK, and 2 are not used. ACK and SEL were also not used by us in the final implementation, although ACK was helpful during debugging. CLK was driven by the processor to synchronize signals. CMD was sent by the processor and DATA was sent by the controller. CMD was sent to the controller through packets. Each packet consists of 9 values that are 8 bits each. We only used 2 sets of commands. The first set forces the controller in analog mode. These commands were obtained from an EECS 373 project website from an earlier semester. The second set of commands asks the controller for button data. The controller responds by putting the data on the DATA line. The DATA line was then decoded in software to determine which buttons were pressed.

Dart Launcher and Project Integration
The darts and launcher we used are standard Lego components (Technic Competition Cannon). The physical components of the project were integrated using K'nex structures. One structure formed the launcher, while the other was a "parking structure" that housed our breadboards and MPC-555.

Software Design:

MPC-555

555 Software

Wireless PS2 Controller
Interfacing to the Playstation 2 controller required us to start by initiating a handshake. This consists of three data commands that must be sent to the controller and must be acknowledged by three data commands sent in return to the processor. After the handshake, the controller returns several bytes of button and analog stick data. All command and data transmissions must be accompanied by a CLK signal generated by the processor. The specification appears to require a 50% duty cycle clock with an 8µs period. We did not have much of an issue with varying the clock speed and settled on a 30µs period to accommodate our other devices. We also wrote a serial interface to read from and write to the controller. Our code also forced the controller into analog mode during initialization so that we could read the analog sticks as well. The commands, button mappings, and overall protocol are well documented in the references.

Distance Sensor and Wireless Link- 555 Side
The distance sensor information was received by the MPC-555 as an integer via its built-in ADC. The integer was then converted to a distance in millimeters. This conversion was done in C by using the information obtained through characterization. Since our characterization was discrete, we decided to output the value as a continuous function by linearizing the value within each step size. Our step size was small enough that the linearization should not have produced noticeable errors. After linearizing, the value was converted to millimeters since our characterization was done in inches. The information was then sent along the serial wireless link through the serial_putchar function given to us by the MPC-555 website.

MPC-823

823 Software

Wireless Serial Link - 823 Side
On the receiving end, the wireless link was connected to a serial port which was plugged into the MPC-823. The software on the 823 would parse this information, removing any garbage caused by the wireless transmission, and display it on the character display. The software on the MPC-823 also determined whether or not the distance value corresponded to Out of Range or Wireless Link Error. We also had to change the baud rate on both ends of the wireless link to 1200 to accommodate the chip's specification.

Character Display
A memory location of 0x2000000 was chosen to interact with the hardware. When we want to write a character, we first checked the busy flag. This was done by a read from that memory location. The hardware was set up so that the busy flag corresponded to bit 31 of the result from the lwz. If the flag was high, then we would continuously loop until the flag went to low. Then we would write to location 0x2000000 the hex bits that the display would use to display the character. Bits 24 to 31 of the word are the data bits and bit 23 is the RS bit. The word would be passed in from a C function as an int. The C function used a very long switch statement to map each character to its corresponding hex value.

Results:

We were very happy with the overall results from our project. All the components worked as expected independently and also after integration. The following are some interesting observations from a design and integration perspective:

Power
We used 7.2V batteries since we did not notice until our final integration was complete that the car motor required a 9.6V battery to operate at full power. The batteries actually charge up to just under 8V, but even at this voltage, the car accelerated very slowly. However, the battery voltage would quickly drop below a usable threshold (for motion on the ground) of about 7.5V. This ultimately allowed us to drive the car for about a minute at best, depending on the slope of the floor.

We also had to use several voltage regulators to power the sensor, servos, and motor at 5 V (the MPC-555 has a built in voltage regulator, so we didn't need to provide one). The reason why we needed multiple regulators is that each regulator apparently only passed up to 1 A reasonably. Initially, when we drew more than an amp from the 5 volt regulator, our car would yield unexpected behavior, such as the dart launcher moving on its own, and the controller losing functionality (it's like the car had a mind of its own!). We were drawing about 2.2 A when fully integrated, not counting the processor. So we used each regulator to provide 5 V and about 0.8 A - 1.0 A for two or three devices. It would have been helpful to reduce the number of circuit boards to make our car lighter, but it was not critical.

MPC-555 Programming
Programming the MPC-555 was frustrating because we often encountered random errors. Often times, these issues were resolved by simply restarting CodeWarrior, but power-cycling the processor was also helpful on occasion. We programmed the processor for the demo by removing the BDM cable (from the processor side) immediately after downloading the code on the processor. This worked fairly well and was much more convenient than programming the flash.

Wireless Link
The wireless link produced erratic behavior beyond a short range. We were restricted by the short range of the signal and the need for a fairly clear line of view. That said, it worked surprisingly well within about 10-12 feet. Since we were not sending mission critical information on the signal, we could afford to, and certainly did, drop some packets.

Distance Sensor
The infrared distance sensor worked well overall, but provided somewhat inconsistent distance readings based on the size of the target that it was pointed towards. When we characterized and tested the distance sensor, we used a rectangular piece of reflective tape. So when the targets differed from the size of that tape, the reading would be a little off.

PS2 Controller
We didn't get around to using the analog sticks in our project. We were able to read them in and get different values depending on the tilt of the sticks, but we decided our time was better invested in system integration. Basically, had we taken an extra hour or two to characterize the analog stick data, they would have been functional. All the other buttons worked flawlessly.

Steering
Due to the nature of the RC car chassis we used, the turning radius on R.O.B. was poor. A chassis with greater steering range would easily solve the problem.

___________________________________

Conclusions:

Overall, our project turned out the way we were hoping it would. We were able to design, build, and run R.O.B. without making many significant changes to our original project proposal. The MPC-555, MPC-823, and Spartan-3 provided more than enough processing power. However, as mentioned above, our final design was low on battery power, which limited R.O.B.'s mobility. If asked to redo the project, we would be sure to check the power specifications of our hardware and get batteries capable of meeting those needs. We would also distribute the work more evenly earlier in the project so that no one person becomes the sole knowledge-base for a component.