Undergrads projects for SS-2016

Available (soon) on olivier.mangin.com/projects/2016-SS.

Project 1: Activity recognition

This project targets recognition of complex human activities from multimodal (vision, language, motion capture) signal.

####Skills:

• programming
• signal (image, sound) processing or computer vision
• Unix

####Tasks:

• setup a perceptual environment
• record a set of human activities along with their linguistic descriptions
• build feature extractors for several modalities

####Potential outcome:

• dataset
• demonstration of activity recognition

Project 2: Is your collaborator better than a tool?

This project studies the components of the decision that leads to asking someone for help in comparison to using a tool.

####Skills:

• psychology
• basic programming
• Unix

####Tasks:

• setup the environment for the recording of human interactions
• run the experiments

####Potential outcome:

• experimental setup
• experiment

Project 2bis: A robot that explores what you can do for it.

This project aims a pilot demonstration of what a robot can learn to get a human doing.

####Skills:

• programming
• Unix

####Tasks:

• setup an interaction space for the robot and the human
• build sensors to detect the state of the interactions
• program basic robot queries

####Potential outcome:

• pilot demonstration
• experimental setup
• experiment

Project 3: A model of attention based on multimodal alignment

This project focuses on modelling attention as the consequence of learning to perceives the environment in a multimodal way.

####Robot:

• Maki or Ergo Jr.

####Skills:

• good programming (python)
• robot control
• Unix

####Tasks:

• setup a robotic platform
• implement visual feature extractor
• setup a speech transcription method
• implement the gaze behavior
• plug in the learning algorithm

####Potential outcome:

• demonstration
• models of attention
• comparative experiments