Modern Robotics

The design challenge is to navigate a robot through a preset course using the knowledge from previous labs and additional research of any kind. Solutions are free from restraint except for the requirement that the voltage source may not exceed 9 volts (standard layout would dictate a 7.2 voltage source). The course layout, dubbed a maze, is a simple square enclosure with 2 barriers protruding from the near and far rails. Black and white tape is laid out inside suggested a course for robots to take or for optic sensors to follow. The interior walls create the challenge while the rest of the course remains very limitless in navigation. Time and accuracy must be taken in consideration, as grade is based on both course time and the robot's ability to maintain consistent time.

The open ended ness of the assignment led to many proposed choices concerning the path of robot, type of control and implantation of chosen design. The most obvious choice was optic sensors, as the tape would ensure a consistent route through the maze and the most accurate times. The design would be as obvious as the route: two sensors controlling the speed or direction of the wheels. When one sensor drifted from the light the wheels would compensate to bring the robot back on track. The idea seemed simple and a sure way to rapidly complete the assignment without trouble. Further thought engendered many concerns: not only must the robot navigate the course but it must also do it faster than the competing teams. Sensors would ensure the robot would cross the finish line, but not with a fast time. The course the sensors must take is loopy and has somewhat sharp turns for the non agile robot. Speed would have to be decreased in order to keep the robot on the track, as a fast and sharp turn could throw the robot off the tape, destroying any possibility of a finish. Another problem arose with sensitivity. The robot, once of the tape a little, would not be able to smoothly get back on the course, resulting in swerving and thus making the course twice as long. With these considerations in mind, we decided that the sensor idea would not be the best choice for our final design.

Our second proposed option gained a notch in the level of thinking, although it was still simple and to the point. We designed a way to use both bumpers to make two turns in order to navigate both walls. The robot would be angled at startup in order to ensure an impact with the far right wall. The robot would then make a left turn and head to the other wall, where it would again impact, turn, and in theory finish the course. The idea seemed consistent at first, and another quick and easy way of completing the course with a decent time. The only necessary circuit setup would be proper wiring of the control boards already constructed on the robot. These circuits would be fail-safe as they were designed by a commercial company and tested before shipping. With the operational status without doubt we could focus on setting the four potentiometers on the board in order to finely tune both turns. Problems began immediately. Not only must both turn be accurate, but the starting angle would also have to be within a very narrow margin or both turn would be thrown off. Various ideas were tried so that the robot could be started at the same angle, but even that did not prevent miss guided turns due to error in the potentiometers. When the idea did work, the robots times were respectable but still not in the range of victory. It seemed indolent to strive for meritocracy, so we opted to decide on another way of designing the robot, although we saw did not discard using bumpers as a way of navigating the course.

Our third design once again gained in complication and achievement. We designed a system of analog turns in order to make two turns at specific times to clear both walls. The turns would not rely on hitting a wall which created two benefits over the previous design: we would not have to worry about starting the robot at a specific point and we would have to rely on the robot hitting the wall, which increasing the length and time of the course. The idea of analog turns had been implemented in previous lab, so no research would be required, only a refreshment of memory. We concluded that two separate timers could be used and then the output would be fed to the timing boards in order for the circuitry to make the two turn. Once again we would have to tune the settings of the four potentiometers, but there would not be as much error in starting angle, as it would be zero degrees. Construction of the circuit was easy and took less than an hour. After tweaking the potentiometers we found that we could easily adjust the timing of the two turn with fairly decent accuracy. After testing the robot several times we noticed a decline in times and more of an error percent in turns. The batteries that ran the timers was draining, thus affecting the entire circuit. Lower voltage meant the wheels would run slower and the timers would act quicker. The result was a robot that turned twice entirely too fast and with little accuracy. New batteries could be used to bring up the speed of the wheels but this meant new settings in all four of our potentiometers on our protoboard as well as the four potentiometers on the robot control board. But once again the power would drain slightly between runs and the times could not be kept constant. The design also left no room for error. Any misturn would throw off the other turn and result in an incomplete course. We decided to discard the idea after many attempts to keep the times consistent and free from tweaking in between runs.

Our next idea kept some of the soul of the previous but did not allow as much room for error. A digital timer would be used to make two turns at precise times. The timing for the turns could be maintained far better than the previous ideas. In order to keep voltage at an ideal level, we used a voltage regulator. The regulator brought down the voltage some but allowed us to more accurately control the robot. The circuit design was used in another course and could almost be exactly transferred with the exception of the timing of the turns. A 555 counter would be used to send out pulses which a binary counter would add. The frequency of the pulses could be set by varying resistance and the timing of the turns could be controlled by the bit output of the counter. We used binary logic to choose the different bits the robot would turn at while making sure that only one combination would make the turn. NAND, OR and NOT gates were used in order to make the turn selection. We were careful to set the frequency low as we did not want the robot to make each turn multiple times. After the binary logic was implementing, analog circuitry was in place in order to feed a signal to the motor control board, which would be used to control the time each wheel varied its speeds.