The CarBot Project Group Members: Chikaod Anyikire, Odi Agenmonmen, Robert Booth, Michael Smith, Reavis Somerville ECE 4006 November 29th 2005 Presentation Outline Project Overview Management Plan Four core tasks Cost Analysis Market Analysis Conclusion 2 Project Overview Objective: To develop an autonomous vehicle that travels from point A to B Features of an autonomous vehicle Detect and react to traffic lights Change lanes to maneuver around obstacles This project will pave the way for significant advancements in transportation and provide a service that will revolutionize the way people travel. 3 Management Plan The management plan consisted of three phases. Planning and Strategy Four Core Tasks Traffic Light Cmucam Robot Integration Robot Coding Final Testing 4 Planning and Strategy Implement a small scale simulation of a autonomous vehicle Preliminary Design Decisions What robot to use as a vehicle? Mekatronics Talrik II What camera to use for detections? CMUcam v1.0 vs. CMUcam v2.0 5 Traffic Light Implementation Determine the best traffic light solution (Bulb vs. LED) COLOR PRODUCED POWER CONSUMPTION REQUIRED VOLTAGE LUMINOUS INTENSITY COST LIGHT BULB LED White Light 20Watts/Bulb True Color at specified wavelength 1 Watts/LED 12 Volts 3-5 Volts 1700 mcd 6000 mcd $1/BULB 2 cents/LED 6 LED Versus Bulb (cont.) LED was the obvious winner in all facets of our requirements. 7 Traffic Light Circuit 8 CMUCam Track the position and size of a colorful or bright object Measure the RGB statistics of an image region Automatically acquire and track the first object it sees On-board real-time vision processing RS232 interface RX – Receive TX - Transmit 9 Camara Integration Motorola 68HC11 Serial Peripheral Interface (SPI) Devices communicate using a master/slave relationship Synchronous serial data link that operates in full duplex MOSI – Master Out Slave In MISO – Master In Slave Out CLK – Clock CMUcam is a SX28 microcontroller Interfaced with a OV6620 Omnivision CMOS camera Serial Port Pin 2 – PC Receive/Scenix Transmit Pin 3 – Scenix Receive/PC Transmit Pin 5 – Ground 10 CMUcam Implementation External Circuit Solution Max3100 to convert the SPI signals to rs232 signals Oscillator to ensure that both the interfaces are operating at the same frequency Software Setup Solution C code needed for the microcontroller to communicate with CMUcam Sets up timing parameters (i.e. baud rate: 9600), etc. Interprets the data from the camera to control robot movement 11 Robot Integration 3 IR detectors 6 IR LEDS ICC Programming environment SPI interfaces with camera Left/Right Motor 12 Stay Straight Algorithm Design Problem: several robot movement problems Solution: error algorithm to prevent robot from pulling left or right Check Global Left / Right Get LA LA = 5 point running average Get RA RA = 5 point running average Compare to LP (initial readings) Compare to RP (initial readings) Convolution Convolution Send Results 13 Change Lanes Is Car in front? YES Check Opposite Lane CAR PRESENT STOP NO CAR PRESENT Change Lanes 14 Traffic Light Detection The cmucam will constantly alternate being tracking colors yellow, red, and green. Several cases that the robot will encounter Yellow Red Green Action Go Stop Track Red / Green Stop Go 15 Final Testing Creating a course to test the autonomous robot Finalizing integration of camera and microcontroller Debugging problems in robot movement algorithms 16 Cost Analysis Estimated costs of car camera system 1/8 size Total non-recurring cost of $980,000 R&D will be the significant portion of initial costs estimated to be $700,000 The return on this investment will be substantial when a working model is produced. The initial investment in this technology will be large because the system has to be reliable, safe, and cost effective. This is an unique technology, and it is difficult to estimate the initial market demand. 17 Market Analysis This product has the potential to revolutionize the way people travel Several cases where automated driving could prevent an impaired driver from operating a vehicle This is a unique product, and being able to introduce it into the automotive industry first will lead to huge technological advantage over possible competitors, and large revenue over time 18 Real World Applications Pedestrian detection Lane following “Blind spot” detection Parking aids The Handicapped Driver assistance – will be interchangeable with the user Military 19 Questions Are there any questions? 20