Huosheng Hu

1
Mike Brady Research SymposiumFrom Oxford AGVs to
Human-Centred Robotics

• Huosheng Hu
• School of Computer Science Electronic
  Engineering
• University of Essex, Colchester CO4 3SQ, U.K.
• Email hhu_at_essex.ac.uk

2
Introduction
4th stage Human-centred environments
3rd stage Unstructured environments
2nd stage Structured environments
1st stage Imaginary world
1920s
1950s
Humans dream of replicating themselves
1980s
Industrial Robots
2000s
Field Robots
Service Robots
Oxford AGV Project
3
Oxford AGV Project (1988-1997)
Tracked vehicle
Robuter
Turtle
LICAR1
OxNav

• Key issues
• Multi-sensor integration
• Sensor-based control
• Autonomous navigation mapping
• Toward unstructured environments

• Funded by EPSRC
• GR/E42416 - 230K (1988-91)
• GR/G37361- 270K (1991-94)
• GR/K39844 - 350K (1994-97)
• plus other grants (Hugh Penny)

Video
4
Human-Centred Robotics

• To operate in human-centred environments that are
  not structured in any way for robots.

• To communicate intelligently with human users to
  know their needs.

• To be adaptive, compliant, fail safe.

• To be easily used by general public who have no
  knowledge on robots programming (elderly
  disabled).

Honda Asimo, 2001
Robotics is the intelligent connection of
perception to action (M. Brady)
5
Typical Applications

Security Robots
Intelligent Appliance
Service Robots
Intelligent Home/ Office/Building
Internet Wireless

Intelligent Transportation System
Intelligent Health Care System
Intelligent Personal Assistant
6
Industrial Robots in Medical

• ROBODOC (1992, IBM), based on an IBM scara robot
  for hip and knee surgeries.

• Rehabilitation Robot

Da Vinci Surgical Robot video
Encoders Force sensors
7
Dextrous Hand for Service Robots
A high-speed robot hand (Komuros Lab at Univ.
of Tokyo) performs impressive acts of dexterity
skillful manipulation (2009).
Shadow C6M Dextrous Hand (20 motors)
IIT Dextrous Hand for the iCub robot (22 DOFs,
extremely compact, light 0.5Kg)
EU-FP7-DEXMART-2009
8
Field Robots for Military
RHex robot
Two robots made by Boston Dynamics
BigDog robot
9
Robotic Fish built at Essex
G6 Robotic fish
G5 Robotic fish
G4 Robotic fish
G8 Robotic Fish
G9 Robotic fish
G14 Robotic fish
Funded by London Aquarium
10
EU FP7 ICT Project SHOAL

• A shoal of robotic fish will
• function in collaboration,
• monitor pollution in a port, and
• map pollution levels its position.

Testing at De Gijon Port, Spain
Sensor networks
Referred from NMRI, Japan
11
Robotic Swarm for Environmental Monitoring

• For optimal coverage of an environmental variable
  for pollution monitoring.
• Control of real and simulated agents using a
  combination of bacteria controller and flocking
  controller.
• Architecture and system design for the System.

Video
11
Prof. H. Hu, School of CSEE, University of Essex,
U.K.
17 June 2014
12
Hands-free Control of Wheelchair
Voice based control
Head gesture based control
13
EMG based control of a wheelchair at Essex

• CyberLink - small wearable headband with 3
  bio-sensors.

Video 1
Video 2
EMG band pass 70- 1000Hz. eyebrow tension EOG
band pass 0.2-3Hz.
Funded by The Royal Society Royal Academy of
Engineering
14
Brain Actuated Control
Video

• 16-CH A/D
• Amplifiers
• Headcap-64
• Electrodes

Funded by EPSRC Partner Oxford University
(Steve Roberts)
15
Multi-sensor Perception
People tracking with service robots

• People are tracked using a multi-sensor solution,
  i.e.
• the camera for face detection
• The SICK laser scanner for legs detection
• Three recognition modalities are integrated
• height
• clothes
• face recognition

16
Bank of Filters for Joint People Tracking
Identification
17
Experiment 1- Tracking 1 Person
18
Multiple Targets

• Data association problem

Different readings
State estimation person A
face / legs of A
State estimation person B
face / legs of B
Different persons
19
Experiment 2 Tracking 3 people
Lux for University Open Day
20
Conclusion

• Opportunity Robot in every home - a personal
  robot era is coming. Like PC Mobile phone, we
  will enjoy these new robots (wearable, portable,
  implantable, pervasive).

21
Future Challenges

• How to effectively conquer home environments.

• Incremental Learning for life-long adaptation

• Memory Machine Consciousness

• Multi-modality human-machine
• interaction

• Interdisciplinary psychology, medical,
• neuroscience, sociology, ethics, etc.

22
Thanks to Mike who led me into the robotics
research at Oxford, which was exciting
enjoyable.

• Thanks to my colleagues and students at Essex
  who made contributions to this talk.

Please see my homepage for more
information http//csee.essex.ac.uk/staff/hhu
Q A