Dr Richard Crowder of the Department of Computing and Electronic Engineering of Southampton University presented a talk on 'Termites, Bees and Robots' to the Isle of Wight branch of Cafe Scientifique on 14 Mar 2016.
2. Presentation will address
Overview of robot development
What are the current developments particularly
at the Biology/Robotics interface
Challenges
2
3. What is a robot…..
I can't define a robot, but I know one when I see
one.
Joseph Engelberger
Developed the first industrial robot in the United States, the Unimate, in the 1950s.
3
6. Robot Timeline
Pre 1800 human like automatons
1801 Jacquard Looms
1920 Capek play: Rossum’s Universal Robots
1940 Master-Slave Systems
1961 Unimate installed in General Motors
1978 Introduction of the PUMA robot
6
7. …..
1997 Soujorner Rover on Mars
1999 AIBO Dog from Sony
2002 Roomba – cleaner from iRobot
2004 -7 DARPA’s Grand Challenges
2013 Google acquires a number of Robotic
companies including Boston Dynamics
7
9. Why Robots
To help humans with tasks that are:
– Dirty
– Dangerous
– Difficult
– Dull
9
10. Where do you find robots
Manufacturing: welding, painting, assemble
Agriculture: crop spaying, harvesting
Medicine: surgery, genome research
Space: exploration, assemble
Military: ordnance disposal, drones
Personal: cars, care assistance
10
11. What is intelligence
…the ability to acquire and apply knowledge
and skills…
Robots can:
– Sense the environment
– Apply the information and skills
– Learn
Intelligence is in the eye of the beholder
11
13. Overview
Developed in the 1950s
Multiple autonomous agents
– Input: light sensors, contact switch
– Output: movement, head light
Artificial life: the robots search the location for
charging batteries (i.e. food when hungry)
Emergent behaviour
No programming
13
15. A Question
How does a termite build its mound and why is
this important to robotics?
15
~1cm
~8m
16. Termites
Very large colonies >106
Co-operative brood care
Division of labour
Self organisation
– Multiple interactions
– Randomness
– Positive feedback
Recruitment and
reinforcement
– Negative feedback
Limited number of
available specific skills
16
17. Key Features
Create spatial-temporal structures:
– Foraging trails, nest architectures, social organisation
Existence of bifurcations when some parameters change:
– Termites move from a non-coordinated to a coordinated
phase only if their density is higher than a threshold value
17
18. Communication
Two types of communication:
– Direct – some form of contact
– Indirect – when one individual modifies the
environment and the others respond
Stigmergy - is a mechanism of indirect coordination
between agents or actions including use of
pheromones
18
19. Constructing the Royal Chamber
19
Queen’s pheromone
determines size of
chamber
Soil pheromone field
that attract workers
20. Robotic System Characteristics
20
Swarm Robotics
Multi-Robotics
Single Robots
Agents
Improved task execution
Improve cooperation
and interaction
Improve robustness
and scalability
Autonomies with learning
and evolutionary capabilities
Complex system
Global communication,
Small number of individuals
Local, distributed control
21. Developing Swarm Systems
Observe a social behaviour
Build a simple model to explain it
Use the model of the social behaviour as a
source of inspiration for solving a practical
problem that has some similarities with the
observed social behaviour
21
22. Why swarms
• The task complexity is too high for a single
robot
• The task is inherently distributed
• Building several resource-bounded robots is
much easier than having a single powerful robot
• The introduction of multiple robots increases
robustness through redundancy
22
23. Characteristics of a Swarm Robot
Robots are autonomous
Robots are situated in the environment and can act
to modify it
Robots’ sensing and communication capabilities
are local
Robots do not have access to centralized control
and/or to global knowledge
Robots cooperate to tackle a given task
23
24. 24
Emergence
An emergent property, e.g. pattern formation, from
more basic constituents
An emergent behaviour can appear as a result of
the interaction of components of the system
Real life example of self-organised behaviour in
humans
– Emergence of paths across grassy area, most
popular paths are reinforced
26. Key features
Mobility
– Wheels, vibrating legs
– Quad-copters
– Fish
Sensors
– Normally quite basic,
proximity and colour
sensors are typical
Communication
– Flashing lights and
multi-coloured LEDs are
common
– Radio systems with very
low ranges
End effector
– Movement of material is
normally by pushing or
simple attraction
systems
26
27. Swarm intelligence…..
Swarm intelligence is an artificial intelligence
technique based around the study of collective
behaviour in decentralized, self-organized
systems
Swarm intelligence systems are typically made
up of a population of simple agents interacting
locally with one another and with their
environment
27
28. …..
Although there is normally no centralised control
structure dictating how individual agents should
behave, local interactions between such agents
often leads to the emergence of global
behaviour.
28
29. Robot foraging
29
• Inspired by Bees
• What is the best approach to maximise collection
• When should we invest effort and time into
collective robot foraging
• What side effects will communication have?
• When is it NOT a good idea for robots to recruit
each other?
32. Individualists: I-Swarm
• Random walk
• Load resource and get its energy efficiency EE
• Bring it back to the base
• Return to the deposit location
• Using odometry
• Neighbourhood search
EE > EEmin
EE ≦EEmin
EE is the energy returned, if greater than EEmin, it is beneficial to return to the original source
32
33. Bee inspired recruitment: B-Swarm
• Depending on initial state a robot can:
• Be a scout looking for a deposit
• Wait in the base for a returning forager
• Can be recruited to another robot’s deposit if it has
higher EE
• Periodically make trips to the base if random walk is
unsuccessful
• Get information from successful returning foragers
33
34. I-Swarm and the environment
• Best performance for 30-75
robots, 100-300 deposits
• Too many robots => physical
interference
• Too many deposits =>
environmental interference
• Too few robots or deposits =>
hard to find anything
34
35. Impact of deposits characteristics
Nectar
• 100 deposits, V=2
• 10 deposit groups of 10
• 3 groups of better quality
Cargo
• 10 deposits, V=20
• Single deposit group of 10
• Uniform deposit quality
35
36. Nectar and cargo
• Nectar: B-Swarms of moderate size rapidly deplete
resource groups, but find new groups hard to locate
• Cargo: Ideal for B-Swarm 36
37. Emergent traffic management
• Congestion around the base created with I-Swarm of 100 when
foraging in an environment with a lot of deposits
37
39. When to forage collectively
1. When resources are hard to find
• Initial collection time is important
• Collection of rare minerals, not picking up litter from
streets
2. When congestion near the base is a problem
• Emergent traffic management
39
40. When to forage individually
1. When resources are abundant
2. When reliability of information is low
3. When it’s cheaper, as extra behaviours are not free
40
41. Summary
Overview of some of my current research into robotics
Many challenges still exist
Biology and Robotics have considerable ability to inform
each over
Intelligence in robots is a significant goal, but requires
contribution from a considerable member of technologies.
41