School of Engineering Department of Electronic and Computer Engineering 171 Development of Bioinspired Tactile Sensor Supervisor: SHEN Yajing / ECE Student: HUANG Zhenghao / ELEC Course: UROP 1100, Fall Touch is an essential manner for humans to receive information from and interact with the world every second of every day. Developing skin-comparable tactile sensors is extremely essential for the fields of robotics and VR, but remains grant challenge. This project aims at developing tactile sensors with force distribution for robotics. If a sensor only have one Hall sensor on the center, we cannot detect the surroundings magnetic change veraciously. Thus, if we have a sensor consists of more segment-sensors, we are able to read more precise data and observe the entire spectrum of magnetic field variations. Based on this technology, robots can perform more delicate tasks as prosthetics, such as mimicking hand joints to hold a pen and write, or applied on prosthetics. Furthermore, humans would have enormous potential in areas like remote communication and medicine if they have such accurate touch sensors. One type of USKIN sensor from XELA Robotics has only one hall-effect-based tactile sensor. Even though the hall sensor is high-accuracy, it would be advanced by adding more hall sensors on it for through detection of force. The hall sensors can be placed as an array to achieve the target. The sensor we aim to create in the project is a 5 hall-sensor-array. We can get more precise data by combining 5 sensors’ information. A grid multiple magnetization approach was proposed, and a soft mXELAagnetic tactile sensor with superresolution was created. In this article, I will introduce a more precise tactile sensor which is composed by five magnetic tactile sensors like mentioned above. This sensor has already attained a certain level of precision, with any degree of pressure creating changes in the internal magnetic field of the sensor, so imitating the perception of skin. We do, however, aim for excellence and expect that the sensors will become even more accurate. We decided to utilize five of these sensors: four around the centre and one in the middle. Development of Bioinspired Tactile Sensor Supervisor: SHEN Yajing / ECE Student: LI Shangqing / ELEC Course: UROP 1100, Spring This paper shows the viability of using machine learning to distinguish when the smart toothbrush is brushing the gingiva versus when it is brushing the teeth, based on audio alone. The paper presented two combined methods of analysing the magnitude data: Statistical and Oscillatory. The model has 85% accuracy for the 34 labelled test cases, which is further verified by predicting the unlabeled random samples, a 90% accuracy. In addition, it is possible to implement a small microphone into the head of a toothbrush, for example SameSky’s CMM-2718-AT-42116-TR has dimension 2.75*1.85*0.95. Therefore, a microphone based smart toothbrush that can distinguish different surfaces is possible to implement.
RkJQdWJsaXNoZXIy NDk5Njg=