Kalman Filter Localization

Introduction to Kalman Filter (1)

• Two measurements

• Weighted least-square

• Finding minimum error

1
• After some calculation and rearrangements if we take wi =
σ i2
Introduction to Kalman Filter (2)
• In Kalman Filter notation
Introduction to Kalman Filter (3)
• Dynamic Prediction (robot moving)
u = velocity
w = noise
• Motion

• Combining fusion and dynamic prediction

Kalman Filter for Robot localization
• Require pose the robot localization problem as a sensor fusion problem
since Kalman filter is an optimal and efficient sensor fusion technique
• Robot localization methods
¾ Markov localization
o the entire perception is used to update each possible robot position
in the belief state individually using the Bayes fomula
¾ Kalman filter localization
o Given a set of possible features, the Kalman filter is used to fuse the
distance estimate from each feature to a matching object in the map.
o The probabilistic update is accomplished by treading the whole,
unimodal, and Gaussian belief state at once.
Kalman filter localization: Robot Position Prediction
• In the first step, the robots position at time step k+1 is predicted based
on its old location (time step k) and its movement due to the control
input u(k):

pˆ ( k + 1 k ) = f ( pˆ ( k k ), u ( k )) f: Odometry function

Σ p ( k + 1 k ) = ∇ p f ⋅ Σ p ( k k ) ⋅ ∇ p f T + ∇u f ⋅ Σu ( k ) ⋅ ∇u f T
 Robot Position Prediction: Example
⎡ ∆sr + ∆sl ∆sr − ∆sl ⎤
⎢ cos( θ + )⎥
⎡ x(k ) ⎤ ⎢
ˆ 2 2 b
⎢ ⎥ ∆sr + ∆sl ∆sr − ∆sl ⎥ Odometry
pˆ ( k + 1 k ) = pˆ ( k k ) + u( k ) = yˆ ( k ) + ⎢ sin(θ + )⎥
⎢ ⎥ ⎢ 2 2 b ⎥
⎢⎣ θˆ ( k ) ⎥⎦ ⎢ ∆sr − ∆sl ⎥
⎢⎣ b ⎥⎦

Σ p ( k + 1 k ) = ∇ p f ⋅ Σ p ( k k ) ⋅ ∇ p f T + ∇u f ⋅ Σu ( k ) ⋅ ∇u f T

⎡ k r ∆s r 0 ⎤
Σu (k ) = ⎢ ⎥
⎣ 0 kl ∆sl ⎦
Kalman filter localization: Observation
• The second step it to obtain the observation Z(k+1) (measurements) from
the robot’s sensors at the new location at time k+1
• The observation usually consists of a set n0 of single observations zj(k+1)
extracted from the different sensors signals. It can represent raw data
scans as well as features like lines, doors or any kind of landmarks.
• The parameters of the targets are usually observed in the sensor frame
{S}.
¾ Therefore the observations have to be transformed to the world frame {W}
or
¾ the measurement prediction have to be transformed to the sensor frame {S}.
¾ This transformation is specified in the function hi (seen later).
Kalman filter localization: Observation: Example
Raw Date of Extracted Lines Extracted Lines
Laser Scanner in Model Space

line j

αj

rj

⎡α j ⎤
R Sensor
z j ( k + 1) = ⎢ ⎥ (robot)
⎣ rj ⎦ frame

⎡σ σαr ⎤
Σ R , j ( k + 1) = ⎢ αα
⎣ σ rα σ rr ⎥⎦ j
Kalman filter localization: Measurement Prediction
(
• In the next step we use the predicted robot position p̂ = k + 1 k )
and the map M(k) to generate multiple predicted observations zt.
• They have to be transformed into the sensor frame
ẑi (k + 1) = hi (zt , p̂ (k + 1 k ))

• We can now define the measurement prediction as the set containing

all ni predicted observations
Ẑ (k + 1) = {ẑi (k + 1)(1 ≤ i ≤ ni )}

• The function hi is mainly the coordinate transformation between the

world frame and the sensor frame
Kalman Filter for Mobile Robot Localization
Measurement Prediction: Example
• For prediction, only the walls that are in
the field of view of the robot are selected.
• This is done by linking the individual
lines to the nodes of the path
Kalman Filter for Mobile Robot Localization
Measurement Prediction: Example
• The generated measurement predictions have to be transformed to the
robot frame {R}

• According to the figure in previous slide the transformation is given by

and its Jacobian by

 5.6.3
Kalman Filter for Mobile Robot Localization
Matching
• Assignment from observations zj(k+1) (gained by the sensors) to the targets zt (stored in
the map)
• For each measurement prediction for which an corresponding observation is found we
calculate the innovation:

and its innovation covariance found by applying the error propagation law:

• The validity of the correspondence between measurement and prediction can e.g. be
evaluated through the Mahalanobis distance:
Kalman Filter for Mobile Robot Localization
Matching: Example
Kalman Filter for Mobile Robot Localization
Estimation: Applying the Kalman Filter
• Kalman filter gain:

• Update of robot’s position estimate:

• The associate variance

Kalman Filter for Mobile Robot Localization
Estimation: 1D Case
• For the one-dimensional case with we can
show that the estimation corresponds to the Kalman filter for one-
dimension presented earlier.
Kalman Filter for Mobile Robot Localization
Estimation: Example
• Kalman filter estimation of the new robot
position pˆ ( k k ) :
¾ By fusing the prediction of robot position
(magenta) with the innovation gained by
the measurements (green) we get the
updated estimate of the robot position
(red)
Autonomous Indoor Navigation (Pygmalion EPFL)
¾ very robust on-the-fly localization
¾ one of the first systems with probabilistic sensor fusion
¾ 47 steps,78 meter length, realistic office environment,
¾ conducted 16 times > 1km overall distance
¾ partially difficult surfaces (laser), partially few vertical edges (vision)
Autonomous Indoor Navigation (Pygmalion EPFL)
Other Localization Methods (not probabilistic)
Localization Base-on Artificial Landmarks
• Landmaks enable reliable and accurate pose estimation by the robot,
which must travel using dead reckoning between the landmarks
• Advantage: a strong formal theory has been developed
• Disadvantage: require significant environmental modification
Other Localization Methods (not probabilistic)
Localization Base on Artificial Landmarks
Other Localization Methods (not probabilistic)
Positioning Beacon Systems: Triangulation
Other Localization Methods (not probabilistic)
Positioning Beacon Systems: Triangulation
• Active ultrasonic beacons
Other Localization Methods (not probabilistic)
Positioning Beacon Systems: Triangulation
• Passive optical beacons
Autonomous Map Building

Starting from an arbitrary initial point, a mobile robot should

be able to autonomously explore the environment with its on
board sensors, gain knowledge about it, interpret the scene,
build an appropriate map and localize itself relative to this
map.

SLAM
The Simultaneous Localization and Mapping Problem
Map Building:
How to Establish a Map
1. By Hand 3. Basic Requirements of a Map:
¾ a way to incorporate newly sensed
information into the existing world model
12 3.5 ¾ information and procedures for estimating
the robot’s position
¾ information to do path planning and other
navigation task (e.g. obstacle avoidance)
2. Automatically: Map Building
• Measure of Quality of a map
The robot learns its environment ¾ topological correctness
predictability
¾ metrical correctness
Motivation:
- by hand: hard and costly • But: Most environments are a mixture of
- dynamically changing environment predictable and unpredictable features
- different look due to different perception → hybrid approach
model-based vs. behaviour-based
Map Building:
The Problems
1. Map Maintaining: Keeping track of 2. Representation and
changes in the environment Reduction of Uncertainty

e.g. disappearing position of robot -> position of wall

cupboard

position of wall -> position of robot

- e.g. measure of belief of each • probability densities for feature positions

environment feature • additional exploration strategies
General Map Building Schematics
Map Representation
• M is a set n of probabilistic feature locations
• Each feature is represented by the covariance matrix Σt and an
associated credibility factor ct

• ct is between 0 and 1 and quantifies the belief in the existence of the

feature in the environment

• a and b define the learning and forgetting rate and ns and nu are the
number of matched and unobserved predictions up to time k,
respectively.
Autonomous Map Building
Stochastic Map Technique
• Stacked system state vector:

• State covariance matrix:

Autonomous Map Building
Example of Feature Based Mapping (EPFL)
Cyclic Environments
Courtesy of Sebastian Thrun

• Small local error accumulate to arbitrary large global errors!

• This is usually irrelevant for navigation
• However, when closing loops, global error does matter
¾ Topology, submaps, combine topology and submaps
Dynamic Environments
• Dynamical changes require continuous mapping

• If extraction of high-level features would be

possible, the mapping in dynamic
environments would become
significantly more straightforward.
¾ e.g. difference between human and wall
¾ Environment modeling is a key factor ?
for robustness
Map Building:
Exploration and Graph Construction
1. Exploration 2. Graph Construction

explore
on stack
already
examined

Where to put the nodes?

• Topology-based: at distinctive locations

- provides correct topology
- must recognize already visited location
• Metric-based: where features disappear or
- backtracking for unexplored openings get visible