UNIGE – Mobile Robots Exam – 2020-2021
Duration: 1 h 30 min
Exercise 1 (under 10 min): Parametrization.
Robot to parametrize
Give the parameter table and the configuration vector.
Exercise 2 (40 min): Robot modeling
Figure 2.1. The robot to study.
We want to establish the equations of the kinematic model of the robot of figure 2.1. In practice, the
robot would typically have at least one more wheel. A simple possibility is a castor wheel without
any actuator, that is why it is completely ignored in this exercise.
Angle matches the usual angle . Angle does not match but it is used because it is more
natural for this robot. So in the table of parameters below, is replaced by its expression as a
funciton of . You will do all the calculations using these angles, which does not make any
difference to you, as the parameter table is given below.
Table of parameters of the robot of figure 2.1.
• 2.1. Give the configuration vector.
• 2.2. Calculate matrices J1, J2, C1 and C2 and C1*.
• 2.3. By considering the determinants of the three minors of C1*, prove that its rank is two,
1/4
except for a certain orientation of the two steerable wheels. Give this orientation.
Recall: the minors of a 2x3 matrix are the three 2x2 matrices obtained by suppressing one
column. The matrix has rank less than 2, if and only if the determinants of the three minors
are zero.
Beware you have to rigorously prove that the rank is not 2 if and only if the values of the
angles are a certain set of values. I insist on the “only if” part. A mere statement like “The
rank is 2 except for ” without a rigorous proof will not be accepted.
• 2.4. Except in this particular wheel orientation, what is the type of the robot?
• 2.5 Assuming the orientation of the steerable wheels are blocked at the above mentioned
orientation (they become fixed wheels), what does the robot type become?
• 2.6. Assuming the robot is never in this particular configuration, give a base of Ker(C1*).
Again, do not just give the base: clearly prove that it is a base.
Exercise 3 (30 min): Observability
The robot below is equipped with a GPS sensor with antenna at point G, at coordinates (0,a) in the
robot frame. The GPS measures the position of the antenna in the absolute reference frame .
G
ym ¯
M xm
The robot and its GPS sensor at point G.
The differential equations below apply to this robot, with x, y and µ the position and orientation of
the robot frame with respect to , and v the speed of the center of the steerable wheel
The goal of the exercise is to study the observability of (x, y, µ). Of course, observability is lost
when the robot is motionless (v=0), but it is not the case we are interested in, so you will ignore
it. We want to find out whether there are other situations where observability is not granted and
what these situations are. These situations must be established based on the calculations required
in the exercise, and by no other method.
1. Establish a suitable state-space model for the observability analysis.
2. Give the expression of a suitable observation equation.
2/4
3. Calculate the gradients of the components of the observation function.
4. Calculate the Lie derivative of the first component of the observation function, and its gradient.
5. Using the gradients calculated above, define a submatrix of the observability matrix and find the
situations (other than v=0) when it is rank deficient (i.e. its rank is less than the dimension of the
state vector) whatever the heading µ of the robot. Analyze the situations you have found. Are they
true observability problems and was the result predictable without calculations?
7. This question is a bonus question, not necessary to get full grade on this exercise. If you don’t
know immediately what to do, your time may be better spent elsewhere.
Now assume the GPS is attached to wheel assembly 1 as in the figure below.
The robot with the new position of the GPS sensor
The coordinates of the GPS in the absolute reference frame are now:
What do you need to change in the state-space representation of the system in order to be able to
study the observability of (x, y, µ).
Attention! The question is not to re-do the observability analysis in this new case, just to say what
needs to be changed in the state-space representation of the system.
Exercise 4 (10 min): Path planning
Two mobile manipulators move in the plane. Their base can move in translation without any
kinematic constraint. The manipulators are RR manipulators with joint limits (no full rotation of
any of the joints).
The environment is made of known fixed obstacles. The nature and shape of the obstacles is not the
object of this exercise, in which we assume we have an efficient procedure to detect the situations
where any of the two mobile manipulators is in collision (base or arm) with either the fixed
obstacles or the other mobile manipulator.
A mission is defined by initial and final configurations of each mobile manipulator. The mobile
manipulators may have to fulfill multiple missions in this same environment.
The figure below shows an example of environment but you must of course not reason on those
3/4
particular positions and shapes of the obstacles. Again, this is just an example.
Your task is to choose a path planning method for this problem and justify your choice.
You have a limited time, so it is up to you to retain the most important aspects of the analysis, that
you consider are enough to justify your choice. Do not get lost in details, you have no time for that.
Ten minutes is more than enough to find the fundamental arguments.
A sample environment, an example initial (top) and final(bottom)
configurations of the mobile manipulators (top)
4/4
