Aid for the Blind

Aids for the perception of disabled people are proliferating with growing intelligent technology. This project aims at adding to the available tools for visually-impaired users, by designing a wearable device that continuously monitors its surroundings and guides the user to safely navigate the streets.

In this project, first a typical test environment with several streets, pedestrian+vehicle traffic, and crossroads must be situated. The designed system must then be able to perform the following tasks within the test environment:

  • Detect and recognize pedestrian crossings (signal displays, line markings, lighting, pedestrian-relevant traffic signs including bicycle road signs, etc.), and notify the user when it is safe to cross the street (by checking the traffic lights, approaching vehicles, etc.).
  • Help the user cross the streets under pedestrian and vehicle traffic with the detected pedestrian crossings and traffic signs.
  • Warn the wearer when obstacles (animate/inanimate, at least 25 cm high) are encountered, and suggest a path to avoid collision with the obstacles.

The device must be

  • Easy and comfortable to use,
  • Light in weight,
  • Free of any hard wiring that may hinder motion,
  • Stingy in energy dissipation,
  • Accurate and efficient in guiding the blind over reasonably complex crossroads, under dense traffic.

Optional feature:

Conversion of text (on boards, paper, etc.) to speech on demand of the user for at least 3 different forms of text, e.g., bus numbers, advertisements, and posters.

The Copycat 

Synchronized motion is a very common behavior both in living things (dancers, schools of fish, or flocks of birds) and in robots (UAV swarms, or industrial robot arms). In this project, a copycat (the agent), which travels in synchronization with a master is to be designed.

A copycat pair consists of a master and an agent, which travel in synchrony. The agent moves at the same speed and with the same orientation as the master. The agent and the master can only communicate in a passive manner, meaning no wired/wireless communication is allowed.

The path the master follows, and its speed profile are both arbitrary and accepted as inputs through a graphical user interface (GUI) on a computer. The agent has no knowledge of the path and strictly follows the movements of the master. Each path is a smooth, natural, human-like track including straight lines and curves. The agent should follow the path and the speed profile as close as possible. The master and the agent should be able to start from arbitrary positions within a radius of 1 m.

The agent should

  • approach the master from a distance, engage and move in synchrony with it. 
  • choose its initial position relative to the master, to start the synchronized motion.
  • be able to catch the master within 5 s after the start.
  • be able to keep its distance from the master at 30 cm (measured from the centers) with a 10% error margin.

Gesture Mimicking Telepresence Robot

The use of telepresence robots for remotely performed tasks is getting popular in various applications such as health care services, operations in high-risk areas, elderly assistance, and so on. The aim of this project is to design and implement a telepresence robot that senses the user’s hand movements and mimics them. The robot is to be designed to replicate the user movements, who writes and draws on an imaginary plane in 3D space. The robot is at a remote location, and will physically reproduce the writing/drawing of the user on a piece of paper.

The system should have the following features:

  • The paper used by the robot can be on an inclined plane, determined by the system designer (can be laid on the ground horizontally, can be hung on a wall vertically, or can be laid on an inclined surface with an arbitrary angle).
  • There will be three pens of different thicknesses and flexibility for the robot to use: (a) a rigid usual thin pen; (b) a long pen of 40 cm coated with thick rubber; (c) a standard pen coated with a long, and thin spiral, as found on wirebound spiral notebooks. The robot should be able to carry these pens by holding them at arbitrary points. 
  • Adequate haptic feedback should be given to the user to have an enhanced sensation of pen-holding compliance that the robot has to have for writing and drawing in its environment.
  • The interface with the user should be natural and intuitive, and should not require any specialized training before use. All hand movements of the user should be executed at speeds natural for a human and should be replicated by the robot in real-time. The user should not slow down for the robot to catch up.

Optional feature: The user can use a VR headset so that other application areas become possible. 

Shadow fixing intelligent canopy

We all have hard times especially at beaches to adjust the canopy as the sun changes orientation and position. Canopies with so many different shapes have been manufactured to avoid this hassle. As an alternative solution, in this project, you are asked to design a highly versatile canopy fixed to the ground by at least 3 poles. The canopy changes its inclination to keep the position and coverage of the shadow as constant as possible with respect to the initial area defined by the poles for different, arbitrary light source positions and orientations changing in 3D. 

  • Each pole should be at least 30 cm long, extending to a maximum of 50 cm, and the distance between adjacent poles should be 50 cm on the ground.
  • Pole positions on the ground are fixed. 

The system should be compatible with a light source, which moves smoothly within a spherical sector of a maximum of ±45 degrees in all directions and assumes at least 3 different positions and orientations in 20 secs.

Training Buddy

Practicing with the best players is critical in advancing your skills in any sport. Table tennis is no exception, yet it is not always possible to train with highly qualified players, since they are often your opponents. In this project, you are asked to design and implement a ping pong ball launcher as a useful training tool with some level of interaction and intelligence.

The ball launcher is a device that 

  • works with verbal commands from the player 
  • launches balls in varying and adjustable swing speeds, serve frequencies, and launching angles
  • exerts topspins and backspins to balls at varying levels (e.g. heavy topspin, slight backspin)
  • determines whether the player was able to hit the ball and records it in a database for sport analytics purposes.

The ball launcher must have five operation modes as follows:

  • Repetition practicing: Consecutive balls are launched in the same style (e.g., topspin to the left).
  • Randomized repetition practicing: Balls are launched as in repetition practicing, but with random variations on parameters such as swing speed, launching angle and landing point.
  • Sequence practicing: Consecutive balls are launched in different styles in a particular order  (e.g., one topspin to the left, one topspin to the right, and then one backspin to the right). 
  • Randomized sequence practicing: Balls are launched as in sequence practicing, but with random variations on parameters such as swing speed, launching angle, and landing point.
  • Game mode: The device simulates an opponent by displaying combinations of all its capabilities and analyzing the skills of the user that require improvement.

The ball launcher must be designed to perform on a real ping pong table. Balls should be launched from at least 1 m away from the net and should land on the opposite side of the table at least 80% of the time in any repetition practicing mode.