Robotic Architecture and Sensors

Sense-Plan-Act Architecture

• Measurements translated into an internal world model

• Generate set of actions to achieve goal

• Generate motion commands

• Shakey’s architecture

Sense-Plan-Act Disadvantages

• Planning in real world takes a long time (Delays in reaction time)

• Sensing not involved in planning (not good in dynamic environments)

• Rely directly on sensed information

Sub Sumption Architecture

Behaviour based architecture, built from layers of interacting behaviours

• Behaviours connect sensors to actuators

• Multiple behaviours can be active

• Higher level behaviours override lower level ones (subsume)

• Higher level behaviours only active when needed

Sub Sumption Advantages

• Led to faster, more reactive robots

• Sense and react in dynamic worlds

Sub Sumption Disadvantages

• Difficult to achieve long term goals

• Difficult to optimise robot behaviours

Hybrid Layer Control Architecture

1. Reactive Layer (Controller) - Behaviour control

   1. Stateless, sensor based, short time scale actions

2. Glue Layer (Sequencer) - Executive control

   1. Has a memory of the past, selects primitive behaviours for controller

3. Planning Layer (Deliberator) - Planning

   1. Plans for the future, time consuming operations

Behavioural Control

• Directly connecting sensors and actuators

• Concerned with here and now

Executive Control

• Interface between (numerical) behavioural control and (symbolic) planning layers

  • Translation of high level plans to low level behaviours

  • Invoking low level behaviours at appropriate times

  • Monitoring execution

  • Concerned with what has just happened and what should happen next

Planning Control

Determining long range activities of the robot based on high level goals

Fully Autonomous

Autonomy level 10

Co operative mission supervision

Autonomy level 9

Mission supervision

Autonomy level 8

Goal supervision

Autonomy level 7

Real time multi robot co operation (work together)

Autonomy level 6

Real time multi robot coordination (swarm robots)

Autonomy level 5

Adapts to faults/events (Compensate for uncertainties)

Autonomy level 4

Robust response to real time faults/events

Autonomy level 3

Changeable mission

Autonomy level 2

Execute preplanned missions

Autonomy level 1

Remotely operated

Autonomy level 0

Sensor Definition

Device that measures some attribute of the world

Transducer Definition

Mechanism that transforms the energy associated with what is being measured into another form of energy

Passive Sensor

Environment provides medium/energy for observation

Active Sensor

Puts out energy into the environment to either change energy or enhance it (sonar)

Active Sensing

Using an end-effector to dynamically position a sensor for a “better look”

Proprioceptive Sensors

Internal Sensors

• Position

• Velocity

• Acceleration

• Motor Torque

Exteroceptive Sensors

External Sensors

• Tactile

• Force + Torque

• Proximity

• Range Finders

• Vision

Inertial Measurement Units (IMUs)

• Position/velocity/acceleration

• Typically contain 3 orthogonal rate-gyroscopes and 3 orthogonal accelerometers

• Measure angular velocity and linear ac eleration

• Estimates position, orientation and usually velocities and accelerations using integration

• Typically include 3 magnetometers to give heading and remove errors

Micro-electromechanical systems (MEMS Gyros)

• Vibrating elements, no rotating parts

• Measures angular velocity by measuring Coriolis force acting on the vibrating mass

• Low power consumption

• Small and low cost

Gyro Drift

• The bias of a rate gyro is the average output from the gyroscope when it is not undergoing any rotation.

• When integrated, this cause an angular error which grows linearly with time

• To correct fuse IMU signals with additional sensors

  • Kalman and particle filters

  • Absolute positioning systems (GPS)

  • Magnetometers

Global Positioning System

Provides 3D position estimate, based on received radio signals from ensemble of satellites

• Compare time delays to triangulate position

• Signal may be unreliable

• Require unobstructed line-of-sight to satellites

• Depends on atmospheric conditions

• Can pass through glass and plastic

• Accuracy 20-25m horizontal, 43m vertical

Sonar / Ultrasonic

Measures time between acoustic pulse and echo

• Low cost, light weight, low power consumption, low computational effort

• Distance = speed of sound /(2 * time of interval)

• Receiver disabled during pulse transmission and after first echo to avoid interference and false readings

• Max detectable distance ~ 6.5m

• Poor directional resolution

• Low sensing rate

• Oblique walls do not produce detectable echos

• Artefacts from multiple reflections