Exam Question Flashcards
How many degrees of freedom does a “‘Synchro Drive” type robot have? Why?
The “‘Syschro drive” robot has a particular configuration: 3 driver and steered wheels but only 2 motors are
used.
The translation motor sets the speeds of the 3 wheel together, and the other motor sets simultaneously
the orientations w.r.t. the vertical axis of each wheel.
So we have that all 3 wheels have at the same time the same orientation and the same velocity.
This configuration, permitted by system of belts, allows to move in any direction of the plane but always
with the same orientation of the chassis (wheels are steered w.r.t. the robot chassis so there is not a way to
re-orienting it)
To conclude, the “synchro drive” configuration on a robot has only 2 DOF.
Illustrate the paradigms for programming robot software architectures
At the system architecture level, we can recognize three main paradigms.
* Hierarchical system architectures organize the five subsystem to support SENSE-PLAN-ACT pattern.
In this paradigm sensor readings are used to build a global world model which is then used to
planning. After the planner has made its decision, it sends instructions to the actuators to achieve
its goal. This is usually not enough for mobile robotics due to the synchronous planning step, even
if we can mate the loop runs faster.
* Reactive system architectures uses only two subsystems (Perception and motor scheme) to support
SENS-ACT patterns. These couplings called behaviors act like reflex so the result allows to take very
fast decisions, focused only in the present or slightly is the past. The result can’t be a global
optimum due to the missing of the planning step, with the risk to getting stuck in a local minimum.
In this paradigm complex behaviors are obtained by combining low levels ones (emergent
behaviors)
* Hybrid deliberative/reactive system architectures organizes the five subsystem (planning,
navigation, cartographer, motor schema, perception) to support “PLAN than SENSE- ACT” pattern.
So using the SENSE-ACT couplings like the reactive paradigm but putting an asynchronous planner
on the top. Sensor reading are used both as input to the behaviors and to build the global map. The
planner coordinates and behaviors decide which one to activate/deactivate. Same implementations
add as intermediate layer (sequencer) to add communication between planner and behaviors.
Illustrate at least 4 sensors useful for indoor robot localization. Explain the basic working
principles.
Bumpers /whiskers: mechanical switchers that notify the robot when it bits an obstacle. Provide
tough local localization information in the form of “bumper x detect an object”
* Proximity sensors: project a ray of infrared light in the environment and measure the intensity of
the ray coming back to detect the presence of a nearly object. They have both an upper and lower
range limit and are sensible to the noise coming from sun rays.
* Ultra sound range sensor: use an ultrasound signal and measure the time of flight of the return
wave to extract obstacle distance. Sensible to cross talk, if more than one is used on the robot, this
can be limited by modulating the output of assigning time slots to each one. Cheap but not very
accurate.
* Laser range sensor: measure the phase shift of a returning modulated laser light w.r.t. the original
signal to measure distance. Since it measures phase shift it suffers from periodicity issues which can
be solved by filtering on the intensity. Can be limited to obtain an area scan instead of a point.
Accurate but expensive
What are exteroceptive sensors in robotics? List at least 5 exteroceptive sensor
Exteroceptive sensor measure the sensing that are external to the organism
Examples are:
* Photoresistor (external light intensity)
* Humidity sensor
* Thermometer (temperature)
* Compass (earth’s magnetic field)
* Inclinometers (heart’s gravitation field)
Define the following terms: Accuracy, precision, Repeatability, Resolution
Accuracy: difference between the sensor’output and the true value.
𝐴𝑐𝑐𝑢𝑟𝑎𝑐𝑦 =
1 − |”measured value”-“𝑡𝑟𝑢𝑒 𝑣𝑎𝑙𝑢𝑒”|
“𝑡𝑟𝑢𝑒 𝑣𝑎𝑙𝑢𝑒”
* Precision: define as range of reading over standard deviation, measure the reproducibility of the
outputs with the same input.
𝑝𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 =
𝑟𝑎𝑛𝑔𝑒
𝜎
* Repeatability: capability of reproducing as output similar measured value over consecutive
measurements of the same constant input quantity. Can only be improved by better components.
* Resolution: maximum variation on the input quantity producing no variation of the measured
output (in absolute value or in percentual % of the input range)
What’s the “Movelt’ package mainly used for?
It is mainly used for controlling the joints of a robotic manipulator, in joint of cartesian space. Performs the
planning and the collision avoidance of user defined obstacle.
What is the difference between automation and autonomy?
Automation is the ability of performing highly repetitive pre-planned tasks under the close world
assumption.
Autonomy is the quality to be self-governing, self-directing freedom and especially moral independence
under the open world assumption.
Illustrate the kinematic configuration of a 3 wheels robot that you know.
The omnidirectional robot has 3 independently actuated fixed Swedish wheels placed on the vertices of an
equilateral triangle, allowing 3 DOF.
This robot controls all 3 DOF just by controlling the speed of the wheels, implying that the degree of
mobility 𝛿𝑚 = 𝐷𝐷𝑂𝐹 = 𝐷𝑂𝐹, making it a holonomic robot.
Which is the difference between a metric and a topologic map?
Metric map saves information about the presence of obstacles, the free space and the unknown space by
storing 2D/3D grid of pixel where each pixel is clusterized in one of the free class above
A topological map saves information about traversable path in the form of a graph, where edges are paths
and nodes are the meeting point between path or landmasts.
Remember that we can go from metric to topological generalizing the internal graph, but not the vice
versa.
5 types of robotic platforms on wheels are:
- one steering wheel and two fixed wheels. The motion of this robot is provided by the two standard
wheels, that are connected one with the other. This configuration has 3 degrees of freedom, and since it
has 3 wheels it has static equilibrium. The degree of maneuverability is 2, while the degree of steerability is
1. - two standard wheels and one castor wheel. This configuration has 3 degrees of freedom, with degree of
maneuverability equal to 1 and degree of steerability equal to 2. In this case we have the castor wheel that
helps us with the balance, but we can add a 4th wheel in order to improve the stability. - three sweedish wheels. This configuration has 3 degrees of freedom, with degree of maneuverability
equal to 0 and degree of steerability equal to 3 - two standard wheels and two steering wheel. In this configuration, the two steering wheel are connected
with the Ackerman steering. The degrees of freedom are 3, with degree of manoeuvrability equal to 2, and
degree of steerability equal to 1. - three spherical wheels. This configuration has 3 degrees of freedom, with degree of steerability equal to
3, and degree of maneuverability.
The student should explain the following terms, in the field of mobile robotics, highlighting the
differences and problems:
* Localization
* Mapping
* exploration
Also, explain what odometry is and why it is not enough to localize a robot.
Localization is the ability for a robot to understand where it is. The robot has to understand where it is
using its sensors and landmarks, but it can happen that the robot is moved without specifying the motion
to it (robot kidnapping), so it has to understand what happened and localize itself again. Another problem
in localization is a wrong reading of the landmarks and the similarities between the environments (like
floors in a hospital).
Mapping is the ability of a robot to build the map of the environment where it is. The problems of mapping
are the fact that the robot has to correct the accumulation of errors, given by the wrong calibration of its
sensors, and the errors in the motions, and it has to understand which are the important features to
memorize in order to be able to recognize them in the future.
Exploration is the ability of a robot to move in autonomy in the (unknown) environment and, eventually,
map it. In this case, for the robot we are in a open world, so it has no knowledge of the environment, and it
has be able to react and correct its motion, according to the obstacles it can find in his path or change its
path if is not able to manage the obstacle.
Odometry is the use of motion sensors to determine the robot’s change in position relative to some known
position. For example, if a robot is traveling in a straight line and if it knows the diameter of its wheels, then
by counting the number of wheel revolutions it can determine how far it has traveled. Odometry is a very
common position sensor for mobile robots, but it has its limitations. Since it is a cumulative measurement,
any sensing error will increase as time passes. Robots may periodically need to use other sensors to
precisely determine the robot’s position to prevent excessive error buildup. Sources of odometry error are:
* Inaccurate wheel diameter measurement
* Different wheel sizes for multiple-wheel drive systems
* Pulse counting errors in systems that use drive shaft encoders
* Slow odometry processing (considering only cumulative counts, not the dynamic count differences)
The student should list the types of locomotion presented in class and illustrate the physical principles
on which they are based and the different energy efficiency.
Locomotion is the act of moving from place to place, is not related only to the body of the robot but to thee
body w.r.t the interaction with the environment. We can have different types of locomotion:
* Crawl: the friction forces is the resistance to motion, and the basic kinematics of motion is the
longitudinal vibration
* Sliding: the friction forces is the resistance to motion, and the basic kinematics of motion is the
transverse vibration
* Running: the loss of kinetic energy is the resistance to motion, and the basic kinematics of motion is
oscillatory movement of a multi-link pendulum
* Jumping: the loss of kinetic energy is the resistance to motion, and the basic kinematics of motion
is oscillatory movement of a multi-link pendulum
* Walking: the gravitational forces is the resistance to motion, and the basic kinematics of motion is
rolling of a polygon.
What are proprioceptive sensors? The student should report at least 3 examples.
Proprioceptive sensors are the sensor that help the robot to understand the internal status of the robot, as
inertial movement unit, wheel encoders, gyroscopes, accelerometers.
The student should describe the fundamental concepts and techniques that refer to the theme of
SLAM (Simultaneous Localization and Mapping).
The SLAM is the process of building the map of the environment of the robot and localizing it in its
environment at the same time. This is done because the localization and the building of the map can be
seen as two faces of the same problem.
SLAM uses probabilistic techniques for building the map, since during the motion in the environment the
robot accumulates errors, since its sensors could be noisy. SLAM with particles filters uses an hypothesis of
a state (particles), and it updates it with the motion done, and it re weights it according to the data
received by the sensors.
The students should explain what the subsumption architecture is and which are the basic elements of
a behavior module.
The subsumption architecture is a purely reactive system based on only two primitives: sense and act highly
coupled into constructs called behaviors. A behavior is represented as an augmented finite state machine
and the principle of functioning is based on the higher levels subsuming the role of lower levels, hence
most important behaviors are executed previously than the other one, hence there is an order of
execution. A behavior is a piece of code executes as a thread that defines a control law to achieving and/or
monitoring a given goal. An arbiter is a very specialized behavior that can decide which behavior should
run. Th arbitration is the resulting spatial or temporal ordering.
