David and Paul Ultrasonic Sensor Investigation


 David Carratu
Paul Gelhaus

Ultrasonic Sensor Investigation
For this investigation of the EV3 Sensors we decided to experiment with the Ultrasonic sensor since we believed that it is easily applicable to autonomous navigation. We wanted to test the tolerances and limitations of this sensor to better understand how the ultrasonic sensor works and in what ways it could be applicable to autonomous navigation. Therefore, we came up with the following questions to investigate:

  • What surfaces does the ultrasonic sensor detect best?
  • How accurate is the sensor’s distance calculation to a wall?
  • What is the maximum detection range of the ultrasonic sensor?
  • Is there an optimal distance from the sensor for object detection?
  • How does the size of an object affect the sensor’s ability to detect it?

We first mounted the Ultrasonic sensor securely to the EV3 LEGO brick to ensure that it would not wobble. We needed it to remain steady so that we would receive clear results. We initially placed the sensor on a platform above the center of the brick, but then discovered that this construction restricted the sensor from picking up on smaller objects on the ground. Therefore, we moved the ultrasonic sensor to one side of the robot. We knew we were going to be using a tape measure to compare the sensor’s readings for distance with actual distance so we constructed a beam in line with the front of the sensor to act as a marker above the tape measure to allows to clearly record the actual distance of the sensor from its object/target. Lastly, on the base of the EV3 brick we created a 2 – wheel chassis and gliders which allowed us to easily push the robot on the ground and not have to pick up the brick every time we wanted to change the distance. This saved us much time during data collection.
To test the limitations and tolerances of the Ultrasonic sensor we completed two experiments which yielded quantitative data and another informal experiment that gave us Qualitative Data. After we constructed our sensor-equipped robot, we tested how well the sensor read the distances to a wide variety of objects. Qualitatively, we found that the sensor is not good at detecting window screens, grass, fabric, and carpet while it detects distance well to surfaces such as the tile-floor, wood, plastic, metal, and glass. These findings make sense since ultrasonic sensors emit ultrasonic sound waves and reads the time differences in the returning sound waves to detect how far away a particular object is. Smooth, hard surfaces reflect sound the best and do not scramble the waves. However, objects without much surface area do not reflect many waves while uneven surfaces scramble the waves. From there we conducted an experiment to see how accurate the sensor is in measuring its distance to a wall. We carried out this experiment by extending a tape measure out perpendicular to the wall and moving the robot back from the wall along the tape measure, which we used to record the actual distance. We started at .25 inches from the wall and recorded the sensor’s data values as we pulled the robot back to 95 inches from the wall. We first noticed that the sensor had limitations in the range that it could detect an object in front of it. We recorded that the sensor can sense an object between .5 and 91.25 inches away. Furthermore, we noticed that at very close distances and at very far distances the sensor was not as accurate as it was at medium distances. This is shown by the fact that our first ten data values and last three data points have much higher % errors than the middle data points. For example, the average % error of the first 9 data points (excluding the first one which is an outlier) is 29% while medium range values (25-50 inches) have an average % error of only .89%. This particular experiment showed us the range of possible detection and also that the optimal distance to detect a large object or wall is between 25 and 50 inches. Our last experiment was finding out the farthest distance that the sensor could sense pieces of paper with different widths. To conduct this, we would hold up the piece of paper right in front of the sensor and move it back until the sensor could not pick up the object. We recorded the value right before detection was lost. From this experiment we learned that the smaller the object, the closer it had to be to the sensor to be detected.
Much of the data that we collected in this investigation shows that ultrasonic sensors are viable options for autonomous navigation, but that they also do have several limitations. As we have proven, ultrasonic sensors, when there is a solid object within their optimal detection range, are very good at accurately detecting the distance to the object. Therefore, they can be used on automatous vehicles as a method to measure the distance the vehicle is from other vehicles, buildings, pedestrians, and any other obstacles that the cars may encounter on the road. However, as shown by the limitations that we discovered such as the range of detection for smaller objects and the need for solid, hard surfaces, ultrasonic sensors cannot be completely relied on as the sole sensors used in autonomous navigation, but can be paired with other types of sensors to provide a cohesive data collection system. 


 Video Description of Robot and our Investigation

1 Comment

  1. Reply David R. Carratu

    I forgot to mention that we did not use LabView to program in this project. Instead we just used the port view on the brick which displayed the sensor value. We inputted the sensor values and our own measurements into Microsoft Excel to create our data tables and graphs.

Leave a Reply

Your email address will not be published. Required fields are marked *