CSCE452 Exam 1 Jason O'Kane

The vague definition of a robot has been compared to how you can’t really tell where ____ begins or ends

Mount Fuji

Four general robot traits

Autonomous, Sensor-driven, Physical, Goal-achieving

The fundamental problem of robotics is

that the robot relies on limited/unreliable hardware to sense and interact with an unpredictable and complex world

Robots exist at the intersection of ____, ____, and ____. Thus, solving any one of these three is not sufficient for robotics development.

computing, acting, and sensing

The four core robotics problems are

Navigation and motion planning

Localization and mapping

Manipulation

Exploration and coverage

Describe the Navigation and motion planning problem

Get from A to B

Describe the Localization and mapping problem

Where am I and what’s around me

explain the Manipulation problem

Grasp, transport, assemble, or disassemble objects in the environment

Explain the exploration and coverage problem

Move to see or touch everything in the environment

An ____ is hardware that contributes to motion

actuator

DC motor

High velocity, low torque, low control

Gearhead motor:

Low velocity, high torque

Stepper motor:

small, discrete rotations provide precise rotation control.

Servo motor:

Can sense and control rotor position, but can’t rotate all the way

Linear actuators:

generate translational motion in various ways

A ____ gathers data from the environment

sensor

Encoders

measure the amount of rotation in a joint or wheel

Infrared sensors:

measure distance by emitting IR light and measuring the intensity of the signal reflected back into the sensor

Ultrasonic/sonar sensors:

measure distance by emitting a pulse of sound and measuring its time-of-flight

Lidar:

measures distance using phase shifts in highly coherent light.

Cameras:

Measure intensity and color of light

RGBD (red-green-blue-depth) sensors:

provide RGB images with an extra channel for distance

Compasses:

Measure direction with earth’s magnetic field

GPS Receivers:

Use satellites to measure absolute position

Inclinometers:

Measure the relative direction of gravity

Inertial measurement units (IMUs):

combine 3 accelerometers and 3 gyroscopes to measure linear and angular acceleration.

What are the two main robot design strats

Brute force your problems and keep adding or improving hardware

Use modest hardware and design your algorithms around those limits

Robot software is often ____, meaning different pieces of code are running on different hardware simultaneously to achieve the desired effect. ROS provides networks

distributed

ROS also provides ____ pieces of code for both reliable, commonly-used techniques, and leading-edge research results.

Modular/reusable

ROS allows code that can be used on ____ robots to work with ____ robots too

real; simulated

A ROS node is an ____

executable program

A ROS topic is a ____ way, ____-to-____ channel from ____ nodes to ____ nodes

one; many-to-many; publisher; subscriber

The ROS Graph is a graph with an edge for each ____

publisher-subscriber relationship

A service is a ____, ____-to-____channel between a ____ and ____

two-way; one-to-one; client; server

(ROS) An ____ is how you ask a node to do something

action

A ROS interface is a ____

data type

A package is a ____

collection of related files and executables

A launch file is a program that ____

launches and configures nodes automatically

A ROS parameter is a ____

bit of configuration data associated with a node.

A ROS bag is a ____

file that stores messages to be replayed later.

The state is the ____. Represented by ____ and ____

collection of all aspects of the robot and the environment that can impact the futurel; X; x

The action (robot modeling, not ROS) is a ____. Represented by ____ and ____

choice the robot makes at a particular time; U; u

A state transition equation describes how ____

actions change the state, and models a future state as a function of a current state and a current action, and optionally a variable theta to represent error. (i.e. x_future = f(x_current, u_current, theta))

The observation is ____. Also a tuple, denoted with ____ and ____

information supplied to the robot at a particular time; Y; y

Ways to model plans (robot modeling)

Drive wheels ____

actively move, powered by an actuator

Steer wheels ____

rotate about the axis perpendicular to the ground to influence direction

Passive wheels are ____

any wheels that don't drive or steer

A differential drive robot has ____

two independent non-steered drive wheels along a common axis.

The state space of a differential drive system is ____

(x, y, 𝜃) for position and direction

The action space of a diff drive system is ____

(vl, vr), the velocities of the two wheels

The ICC of a diff drive system is what we call the point where ____

the robot moves about with a constant angular velocity

l in the diff drive equations represent ____

the distance between the wheels

R in the diff drive equations represent _____

Distance from ICC to robot center

Location of ICC formula

C = (x - Rsin(𝜃), y + Rcos(𝜃))

When multiplying matrices, each row of the product matches matrix ____ (A or B) and each column matches matrix ____ (A or B)

A, B

What are the two special cases for differential drive?

If vl = -vr, R=0, i.e., the robot rotates in place

if vl=vr, R is infinite, robot moves forward

A diff drive is a nonholonomic system because it ____

cannot move directly sideways.

How do you find the ICC location for a non differential drive system? What are the four cases and what do they show?

the ICC is the intersection of all the lines perpendicular to the rolling direction of all the wheels

If the lines intersect at a single point, that point is the ICC.

If the lines overlap, the ICC can lie anywhere along that line.

If the lines are distinct and parallel, the ICC is 'at infinity' in that direction.

If the lines do not share any common intersection point and are not all parallel, the robot cannot move.

Bicycle drive has a ____

steered wheel in the front and a non-steered wheel in the back.

Tricycle drive has a ____

steered wheel in the front and two non steered wheels in the back

A synchronous drive robot has ____

All the wheels always ____ and ____.

three steerable drive wheels

point in the same direction

drive at the same speed

Ackerman steering has ____

two steer wheels

What is special about the inner steer wheel in an ackerman steering system? Why is this necessary?

the inside wheel has to turn more than the outer wheel. If they didn’t, i.e. if their rolling directions were always parallel, then the lines perpendicular to their rolling directions would also be parallel, meaning the robot cannot move.

What are the four alternatives to wheels?

Legs

Legs and wheels

Legs that rotate like wheels

Hybrid terrestrial/aquatic locomotion

Describe the bug0 algorithm. Does it always find a solution?

Go directly to the goal

If you hit an obstacle, turn left and wall follow until you can once again go directly to the goal

Repeat

Bug0 is incomplete, it can trap you in an infinite loop

Describe Bug1

Go directly to the goal

If you hit an obstacle, turn left and wall follow, circling the entire obstacle, and recording your distance from the goal the whole way

After returning back to where you first hit the obstacle, wall follow the same obstacle back to the point along the obstacle that was closest to the goal (use the shorter direction left vs right)

Repeat

Describe Bug2

Record a straight line between the start and goal states

Go directly to the goal

If you hit an obstacle, turn left and wall follow until you hit the line

If you are now closer to the goal than when you hit the obstacle, go directly to the goal, else keep following the obstacle

The ____ is where you hit an obstacle, the ____ is where you return to goal seeking

hit point; leave point

What do the variables D, M, p_i, and n_i represent in the bug algorithm formulas represent?

D - upper bound on path length

M - number of obstacles

p_i - perimeter of the ith obstacle

n_i - number of intersection points between the start-to-goal line and the perimeter of the ith obstacle

If each obstacle is a polygon, the visibility graph is a weighted graph with a set of nodes that includes the ____, ____, and all ____

start-state; end-state; obstacle vertices

In a visibility graph, we Include all edges that don’t ____

go through an obstacle

In a visbility graph, for any edge (u, v), it is considered tangent if the infinite line which (u, v) is a segment of does not ____

intersect either polygon that u and v are a part of.

The reduced visibility graph only has ____ edges.

Tangent

How do we know that the shortest path must be along the visibility graph?

Proof by contradiction. Say the shortest path contains some segment that is not an edge in the visibility graph. This means the segment changes direction at a point that is not a node in the graph (i.e. a point in free space or along an edge). This means that the segments leading to and from the point could be connected with a straight line. By reducing two non parallel segments to one straight line, the path has been made shorter. Therefore this cannot be the shortest path. Hence proved.

Does the visibility graph method of finding shortest path apply to 3D navigation problems?

No

Why is the visibility graph method of finding shortest path considered dangerous?

In the real world, having a robot come so close to an obstacle edge or vertex could cause a collision