Titulo Estágio
Object Classification with Electronic Skin for embodied Intelligence
Áreas de especialidade
Sistemas Inteligentes
Local do Estágio
Department of Electrical Engineering, University of Coimbra
Enquadramento
An electronic skin is a soft ultra-thin film populated with a large number of sensors integrated. Such system will be used for object classification.
This is an excellent project for students who like the field of classification, such as SVM, Neutal Networks etc. Application with a prosthetic hand and a robotic arm will be explored.
Students will recieve a prototype of e-skin and the data acquisition boards for this purpose.
[url=https://sites.google.com/site/fsrlabisr/msc-projects]Project 2[/url]
Objetivo
Student will receive an electronic skin that is composed of over 25 measuring electrodes. The hardware is in form of a stretchable patch from silicone.
Each sensor provides information on proximity of a conductive object or the human hand to the soft skin. When different objects approach the skin, e-skin provides 25 values from 25 capacitors. The platform is also able to measure pressure, which can be used for classification of a non-conductive objects.
In summary, this is a e-skin that replaces vision.
The project consist of making a classification algorithm (e.g. SVM) which is able to "guess" what object is approaching. This requires acquisition of data of different objects to train the classifier.
The e-skin can be then demonstrated for different applications. For instance, a working prosthetic hand which was developed by our group can be used in conjunction with the e-skin, in order to detect the object and autonomously close the hand based on the object which is approaching.
Plano de Trabalhos - Semestre 1
1. Get familiar with the hardware
2. Acquire Data
3. Develop an early version of the classifier
Plano de Trabalhos - Semestre 2
4. Train the system with multiple objects
5. Classifier should classify different objects
6. Test, improve and validate the classifier
7. Demonstrate an application for autonomous grasping with an existing prosthetic hand
Condições
Student should have some experience on Machine Learning and Classification methods.
He will be welcome to choose a co-supervisor from the Informatic department.
Observações
Laboratory of Robotics and Mechanism of ISR-UC has close collaborations with the Carnegie Mellon University in the areas related with soft electronics, soft hands, Tactile sensing, Human Machine Interaction, novel interactive concepts, and is constantly looking for talented students to join the team.
In the Laboratory you find an excellent range of prototyping equipment, including four different 3D printers, Laser cutting Machine, Polymer processing equipment for casting sculptures and sensors, a wide range of electronics hardware for human machine interaction.
Students are welcome to visit the lab and the hardware before making a decision. Just send an email to mahmoud@isr.uc.pt
Also since the proposals are advertised also for Erasmus Students, students are advised to talk to us and also to select more than one project from our list as option 2 and 3.
https://sites.google.com/site/fsrlabisr/home
Orientador
Mahmoud Tavakoli
mahmoud@isr.uc.pt 📩