Lab 11

Date: 25/11 - 2010
Duration of activity: 14 - 17
Group members participating: Morten Rasmussen, Kenneth Sejdenfaden Bøgh, Morten Nikolaj Pløger.


Goal:


Today we will try to figure out what end course project we would like to do.


Results:


list of projects considered with a short description of each, e.g. a description could be:"robot that can dry wet spots on the floor during a handball match".

The list below is ordered. That is we would prefer to do project number 1. The reason for the choice of this project is that we think it is doable and yet contains enough challenges. The other projects are either too difficult or too trivial from our point of view. It should also be mentioned that although we are only three participants today we will be four people working on this project. As such we believe that we will be able to take on this rather large task.  


Projects considered:

1. Laundry pickup robot: two robots. One drives around and picks up laundry. Another is stationary and raises a platform to put the laundry into the laundry basket.

Sensors needed: 3x Sonic Sensors, 3+x Pressure Sensors.

Here were would need two NXT’s, for two individual robots, one to retrieve the laundry and deliver it to the second robot which then puts the laundry into the laundry basket. The laundry retrieving robot would be a Behaviour based robot, with behaviours such as Wander(looking for laundry), Pick up laundry, Deliver laundry to laundry basket, avoid obstacles, recharge. The second robot would be mounted to the laundry basket with a mechanism which enables it to lift clothes. The scenario is: First the wandering robot delivers a piece of cloth to the mounted lifting robot and leaves the clothes on a platform. Then the mounted lifting robot lifts up the platform to the edge of the basket, where it will tilt the platform so taht the clothes fall into the basket and returns back to its initial position.

Some thoughts about the behaviours:

Wander will require some sort of pressure sensor attached to the robot in such a way that the sensor will sense laundry. This could be achieved using multiple pressure sensors in front of the robot in a line with a flat panel attached close to the ground though not touching the ground (multiple sensors can in this case be connected to the same port).

Pick up laundry is invoked after Wander has detected some clothes on the floor. The way we imagine the robot would be able to move laundry would be to use two ‘hands’ in the form of pads with rough surfaces, which it then move together around the clothes and start dragging it by moving backwards towards the laundry basket.

Delivering laundry to laundry basket involves three Sonic Sensors. The laundry retrieving robot sends a signal to the other basket robot ‘Hey I got laundry!’ and the basket robot starts sending a constant signal with its Sonic Sensor. Now, the retrieving robot can find its way to the basket in a Braitenberg[1] fashion, hence feeding the input directly to the motors, since it is equipped with two Sonic Sensor’s. When the laundry is delivered the basket robot stops sending the ultrasonic signal.

Avoid obstacles would make use of the two Sonic Sensor’s. This could lead to some interesting problems since we risk disturbance between the two sensors and hence be forced to make e.g. a distribution in time to make atomic measurements. This behaviour also includes a Pressure Sensor at the back of the robot with the sole purpose of avoiding objects while dragging laundry since the Sonic Sensor’s at this time are unable to do nothing but listen in order to locate the basket.

We expect to be able to present a robot that is able to detect and retrieve laundry to an active landmark in the form of an NXT, avoiding objects both as it is searching and on its way back to delivery.


2. Multicolour printer: One robot that is a able to draw in several colors and print line-by-line like an actual printer does. ie something like this http://www.youtube.com/watch?v=zX09WnGU6ZY but using nxt

This would require a single NXT and a lego setup that would be able to feed the paper in a controlled manor.
One motor will feed the paper forward, another one will control the left to right motion and the last one will control which pen is currently used to draw.
The software would be client server based. That is the NXT would be the server and then PC’s would be able to connect to it using bluetooth. The client software would then need to transform the bitmap picture to a resolution that the robot would be able to print.

The most difficult part of this project would be the transformation of the pictures and to get the tacko counter to move the paper exactly the same distance every time(so that all pixels have the same width)
Also it might be hard to avoid the pen “bleeding” if it stops on the paper for a short while.

At the end we would be able to present a printer that could print bitmap pictures, although they would probably be in a rather crude resolution(ie pixilated).


3. Stalker robot: robot that picks out a moving target and follows it around stopping when the target stop and moving when the target is moving

For this project we considered using the sonic sensor to keep track of the stalking victim. The idea was to make the robot follow a person and keep following this person for a given time period or until the robot looses track of the victim. We discussed how the software should interact with the world. This should be done through a behaviour model, where the behaviour should be in the layer of: find victim, follow victim.

This project would, in our opinion, be either too difficult or not difficult enough.
The part where this project would be too easy, is when the robot is just stalking a random victim, which may change, for instance if the person enters a group the robot may stalk any random victim of the group.
The part where this project would be too difficult, is the opposite situation as above. The robot initially stalks a random victim and keeps stalking this victim, if the victim enters a group, the robot should stalk the same person when the victim leaves the group.

If we chose this project, we should be able to present a robot, which would be able to stalk a random victim, but if the victim enters a group, the robot would choose a new random victim to stalk.



4. Hamster ball:  Pet robot, something similar to http://xkcd.com/413/

For this project we would need two large hamster balls and two nxts. It would require some modifications though. The camera on top in the comic is too advanced to build for this course, so we would be dropping that. Since real hamster balls are not very see through we would not be able to put light sensors inside the robots.
Instead we would create two pets and and have them use bluetooth to locate each other and stuff like food, “litter tray” and bed.  They would then have an amount of behaviors to make them act like real pets. As such they would wonder around until they found another pet in which case they would start playing, and go to the food when they become hungry, go to the sleep area then they are running out of power etc.
The idea is, although we do not know if it will work, to connect to the bluetooth device which is named according to what it represents. Ie a phone or another bluetooth emitting device acts as food by having the name “food” and another acts as sleeping area etc.
After the connection is established the robot will move to that object by measuring the strength of the bluetooth signal.

There are two major obstacles for this project. First we need to find a hamster ball big enough to contain the lego construction, besides this it also needs to be able to open in two half's. Most hamster balls have a small hole that opens instead of being in two half's, as described above, which will not be big enough to fit the nxt through.
Secondly the idea of using bluetooth strength to navigate to a point might not work since the signal strength measurements might not be accurate enough to determine a direction. Even if it works it might be difficult to implement efficiently.

Assuming we can find the hamster balls we would be able to present two robots that can at least wonder around a play very simply with each other. The scenario is: both robots look for the other robots bluetooth signal. Once a bluetooth signal is found, one robot sends a small sound file to the other using bluetooths. The robot that received the file then starts playing it, while the robot that originally send the file starts to listen for the specific sound sequence. This would enable the robots to identify one and another.


References:

[1], Braitenberg, V. 1984. Vehicles, Experiments in Synthetic Psychology London, Cambridge: The MIT Press.