Lesson 11

Lesson 11 description

List of projects

Date: 25. November 2010

Duration of activity: 13 - 15

Group members participating: Nikki, Amos & Knud

---

GOALS for lesson 11:
Our goal is to choose and make a initial description of the end course project

List of projects considered:

Sex bots/Embodied evolution

Project description

Robots that are able to “breed” and evolve.

Breeding is accomplished by exchanging artificial genes over IR, the robots “die” and their children with the new genes take over the empty robot shell. A percentage of mutation occurs when breeding, to enable evolution of new behaviours.
If a robot is not able to find a mate within its predetermined lifespan its genes will go extinct.
The behaviors of the robots could be implemented with inspiration from the braitenberg vehicles.

Hardware requirements

• Numbers of NXTs: at least 2
• Sensors: IR Link, ultrasonic sensor (or touch sensor), light sensors
• Actuators: The motor

Challenges

• Getting the basic behaviours working, without genes.
• Making the behaviors reproduce according to “survival of the fittest” in order so that the genes appropriate for their survival are inherited and fostered naturally or automatic. Exchange of genes.

Final outcome to be presented

• Embodied agents that evolve and adapt, showing different/optimal behaviours according to the internal states and different stimuli's from the “world” .

Synthetic Creatures

Project description

• Build a robot that can behave like a creature in an environment with elements that the robot can react to. Example from light.
• The robot can be inspired by the braitenberg vehicles with different behavior/reactions from different elements in the environment.

Hardware requirements

• Numbers of NXTs: at least 1
• Sensors: Light Sensor, ultra sonic sensor
• Actuators: The motors

Challenges

Making an environment that the creatures can interact with/in, making the behaviors/reaction work properly.

Final outcome to be presented

A environment where the creatures live and interact with each other and the environment, with light and food sources.

Navigation through LEGO road elements

Project description

• The project is to construct a LEGO car that can drive on the LEGO roads, and to build a map of the LEGO road network. Afterwards the car should be able to get from one known landmark to another by the shortest known route.

Hardware requirements

• Numbers of NXTs: 1
• Sensors: Ultrasonic sensor, color sensor
• Actuators: The Motors

Challenges

Recognise landmarks, calculate the shortest route between various landmarks, determine arrival at the different landmarks.

Final outcome to be presented

The Lego robot mapping landmarks, driving the shortest routes to different landmarks on the lego roads

Selected project:

We have chosen Sex bots/Embodied evolution. This is what interests us the most, and we are able to borrow 4+ NXT’s.

We want to create autonomous robots able to adapt by embodied evolution. We will base it on a simplified version of the Motivation Network described in “Motivation Networks - A Biological Model for Autonomous Agent Control” article [Krink].

The robots environment will consist of a walled of arena of 3x1½ meters, with a food source in one corner. the food source should consist of a green colored floor/surface and a light source. This enables to robots to find their way to the food source by using a light sensor, and detect when it is on top of the food by using a color sensor.
Agents energy level decrease over time, if it reached 0 the robot dies, and its genes go extinct. To replenish energy the agent needs to eat.
Agents will mate when they have reached a set maturity age and their energy levels are sufficient. When mating, the two agents “die” and is “replaced” by their offspring, a mix of the agents set of genes with a certain percentage of mutation. Genes determine when different behaviours will be active, as described in the article [Krink]. We will have to research different options for mate detection and mating (some form of communication to exchange of genes between the two agents).

In the article [Krink], food sources diminish when the agents eat and regenerates over time, so the behaviour of the agents directly affects the environment, also predators will hunt and kill the agents.
In this project we will initially not make diminishing/regenerating food sources, and we will not introduce predators. If time permits it we will try to implement predators and/or diminishing/regenerating food sources to see if behaviours change.
Because action affects stimuli, and stimuli affect actions we still hope to see emergent behaviour even though the agents wont affect the environment.

Initially we want a range of basic behaviours and internal states.

Internal states:

• Energy level
• Sex drive
• Age

Mapping functions:

Artificial genes:

Motivation variables:

• Feeding
• Food searching
• Breeding

Basic behaviours:

• Feeding
• Food searching
• Find mate
• Breeding

If time permits, we will expand this with behaviours such as resting and flight (flee from predators).

Conclusion:

Work plan:

We have agreed to work at least each Thursday. The plan is as follows:

Project plan:
First Lab Session - Making the basic behaviours of the robots.

Second Lab Session - Starting on a monitor program and looking into genome based behavioural states of the robots.

Third Lab Session - Look at IR-communications for the robots to interact (switch genomes) and testing genome evolution.

2/12-2010 Goals:

• Finish this labreport and create the first project labreport.
• Create some of the basic behaviours of the robots in, with only one behaviour active at a time.
• Avoid clash - Avoid running into walls and other robots
• Find food - Find food source by going towards a light source
• Eat food - Detect food (by colour) and show visually that the robot is eating

9/12-2010 Goals:

• Figure out a way to detect a mate and do the mating (exchange genes/code)
• Start on framework for Monitor/Server program

16/12-2010 Goals:

• Continue framework for Monitor/Server program

• Implement the chosen mate detection and mating (communication part)
• Research and start implementing a Motivation Network

References:

Krink:
Thiemo Krink (in prep.), Motivation Networks - A Biological Model for Autonomous Agent Control
(http://legolab.daimi.au.dk/DigitalControl.dir/Krink.pdf)