The Trouble Quad
Design Constraint Analysis and Component Selection Rationale
Author: Tyson Mowery
Introduction:
Our team’s design project is to design a remote-controlled vehicle that, given a set of waypoints, can find its location on an unknown map and traverse those waypoints while mapping out obstacles it comes across. The vehicle will retrieve its waypoints from a web-based application via a Wi-Fi network card onboard. Once its destination coordinates are received it will compute the best route and visit those points by controlling the two motors/wheels centered on each side. In order to verify that it has located the correct point on the map, the vehicle will have a camera directed towards the ground beneath it, and the image from this camera will be processed to locate special pictured squares to be place on the floor. Each of these squares would have some quality that uniquely distinguishes it from other similar squares so the car can place itself correctly on the map and continue in the right direction.
Our “smart” car design has several constraints due to the fact that all of the components must be located on a mobile, PC-controlled car. The vehicle must be mobile and remain versatile despite the mounted components. Speed however is a non-issue for this design. Another major concern is power consumption. The Rabbit controller is likely to use a lot of power so it is essential to reduce the power consumed by other components to a minimum.
Constraint Analysis:
Size, Weight, and Cost Constraints:
The chassis building materials come in 12 inch by 12 inch sheets so assuming there is only a single “deck” on our vehicle, all of the components need to fit on this sheet. The design needs to be light enough that the small DC motors can drive the vehicle and heavy enough to remain firmly on the ground (1-5 pounds should suffice). We set our total budget at $1000. A cost constraint breakdown is as follows:
$600 for the processor and Wi-Fi,
$100 for the CMU camera,
$50 for the infrared sensor, and
$250 for car construction.
Our design involves four major components that will be discussed individually in this section concerning other constraint issues.
Microprocessor:
Our project requires a processor that can handle a heavy workload and has a large memory. It also has to allow for connectivity to a CMU camera as well as an infrared sensor. The processor will have to do a great deal of image processing which is quite processor intensive and while time is not a substantial issue, for testing purposes and real-world application, a fairly quick processor speed is necessary. Along with image processing, it will need to solve best-route algorithms before departing from its starting point. Fortunately, these two main functions will not be performed simultaneously. So a Rabbit processor will almost certainly be needed to handle the computational requirements. Another significant design requirement is for it to have Wi-Fi connectivity. This will control all of the vehicle functionality so power consumption need not be limited for this element given its importance.
Sensor:
There will be a sensor attached to the front of the vehicle used to detect obstacles that might fall into its path. The sensor must be able to detect objects from a reasonable distance (10 to 100 centimeters) and in all possible lighting environments. Since the detection distance and power consumption are directly related, it is a good opportunity to reduce power consumption since the required distance is minimal. Due to the lighting issue, an infrared sensor will be used to ensure that objects can be detected in the dark. Otherwise, this component has no substantial constraints.
Motors:
The motors must meet three criteria. They must generate enough torque to turn the wheels and move the car. However, considering the car will be roughly the size of a standard remote-controlled car, any motor should successfully turn the tires. The motors must be reversible in order to perform in-place turns. Finally, there is a need to have some manner of feedback capability that can relay information back to processor regarding distance traveled. Because speed is not a concern for this project, RPM ratings are negligible. The processor will be using pulse-width-modulation to control the motors so if its range is not great enough, a voltage amplifier may be necessary.
Camera:
The camera has to be easily mounted under the car so it should be relatively small. It will be identifying striped pictures from a very close distance so resolution is not a priority. The final quality the camera must possess is a connection that corresponds to a connection on the microprocessor. Almost all CMU cameras have the same resolution and 176x255 should easily satisfy our needs.
Selection Rationale:
Microprocessor:
Our team had two paths we could take as far as choosing a microprocessor. One route was to buy each part independently and attach them ourselves. This would involve a main board with an ethernet connection and a 802.11 Wi-Fi card as well as the adapters to attach them. Individually this package would cost approximately $400. The other option which our group is opting for is a Wi-Fi application kit from This includes all of the above mentioned hardware integrated onto one board as well as some necessary software built in (TCP/IP stack). Taking this path would save us a great deal of time and energy. It also makes the physical vehicle design a bit more straight-forward by having only one, smaller component to be mounted, but it does however come at the significant cost of $600.
Sensor:
The cheapest option of $12.50 was an infrared sensor that could detect from 4 cm to 30 cm and reported this reading as an analog voltage between 0 and 5 volts. Another, similar version reports the distance digitally instead and only takes readings when requested as opposed to a continuous reading. This would mean less power consumption. Also the range is greater, 10 cm to 80 cm, and it costs $19.00. Another option is to use a high performance ultrasonic range finder. However, given our project, this high-end equipment would be simply unnecessary and is certainly not worth $34.50 when considering our needs. Our group opted to use the $19.00 sensor. Although more expensive than the other infrared option, this better suits our needs. The digital output would be more reliable and accurate and the added range has its benefits as well. The sooner an obstacle could be identified, the more efficiently the car could determine its best route.
Motors:
For the motors, our group needed something that would perform effectively, but not cost too much in terms of price or power. Two options from were considered, a Pittman geared motor with optical encoder or the version without an encoder. The dilemma here is whether or not a decoder is necessary for our design. It was decided that while not absolutely essential, it could provide some very helpful information and more accurately control our vehicle as opposed to deriving our own method for determining the traveled distance based on the provided motor input voltage and extensive testing. Other attributes such as size, weight, and torque were all relatively equal so our group chose the GM9236 with optical encoder.
CMU Camera:
offers two different CMUCams; the first captures pictures at a rate of 17 frames per second ($109 CMUCam) while another can operate at 50 frames per second ($199 CMUCam2). Other advantages of the second version involve being able to control up to five servos and the ability to directly gather information about the image such as mean color variance and a histogram of each color channel. However, these advantages are unnecessary for our application. Both cameras have the same resolution and considering the operational speed of our vehicle, the frame rates are somewhat irrelevant. Our camera will be stationary so the tracking capabilities of CMUCam2 would go unused. The obvious choice was to use the cheaper camera.
PART / VENDOR / PART NUMBER / COST / QUANTITY / TOTALMicroprocessor / RCM3100 Rabbit WiFi Application Kit / / N/A / 599 / 1 / 599
Object Sensor / Sharp GP2D02 IR Ranger / / R19-IR02 / 19 / 2 / 38
CMU Camera / CMUCam / / R140-CMUCAM-KIT / 109 / 1 / 109
Motors / Pittman Geared Motor w/ Opt. Encoder / / GM9236C534-R2 / 49 / 2 / 98