Learning and Inferring Transportation Routines

1
Learning and Inferring Transportation Routines

• By
• Lin Liao, Dieter Fox and Henry Kautz
• Best Paper award AAAI04

2
AIM of the paper

• Describe a system that creates a probabilistic
  model of a users daily movements through the
  community using unsupervised learning from raw
  GPS data.

3
What this probabilistic model can do?

• Infer locations of usual goal like home or work
  place.
• Infer mode of transportation
• Predict future movements (short and long-term)
• Infer flawed behavior or broken routine
• Robustly track and predict behavior even in the
  presence of total loss of GPS signal.

4
Describing the model

• Hierarchical activity model of a user from a data
  collected from a wearable GPS.
• Represented by a Dynamic Bayesian network
• Inference performed by Rao-Blackwellised particle
  filtering

5
gk
Goal g
gk-1
fgk
Goal switching fg
tk
tk-1
Trip segment t
ftk
Trip switching ft
mk-1
mk
Transportation mode m
Mode switching fm
tk
tk-1
fmk
xk
xk-1
x Location, velocity and car
Tk-1
Tk
zk-1
zk
GPS reading z
6
Location and Transportation modes

• Xk gives location, velocity of the
  person and location of persons car
• Location lk is estimated on a graph structure
  representing a street map using the parameter ?k.
• zk is generated by person carrying GPS data.
• mk can be Bus,Foot,Car,Building
• t models the decision a person makes when moving
  over a vertex in the graph, for example, to turn
  right on a signal.

7
Trip segments

• tk is defined by
• Start location tsk
• End location tek and
• Mode of transportation tmk
• Switching nodes
• Handle transfer between modes and trip segments.

8
Goals

• A goal represents the current target location of
  the person.
• E.g. Home, grocery store, locations of friends
• Assumption Goal of a person can only change when
  the person reaches the end of a trip segment
  level.

9
Inference

• Inference estimate current state distribution
  given all past readings
• Particle filtering
• Evolve approximation to state distribution using
  samples (particles)
• Supports multi-modal distributions
• Supports discrete variables (e.g. mode)
• Rao-Blackwellisation
• Particles include distributions over variables,
  not just single samples
• Improved accuracy with fewer particles (hopefully)

10
Types of Inference

• Goal and trip segment estimation
• GPS based tracking on street maps
• Estimate a persons location by a graph-structure
  S (V,E)
• Aim Find the posterior probability by
  Rao-Blackwellised particle filtering.

Prior by Kalman-filtering
11
Learning

• Structural learning
• Searches for significant locations, e.g. user
  goals and mode transfer locations
• Parameter learning
• Estimate transition probabilities
• Transitions between blocks
• Transitions between modes

12
Structural learning

• Finding goals
• Locations where a person spends extended period
  of time
• Finding mode transfer locations
• Estimate mode transition probabilities for each
  street
• E.g. bus stops and parking lots are those
  locations where the mode transition probabilities
  exceed a certain threshold

13
Detection of abnormal behavior

• If person always repeats usual activities,
  activity tracking can be done with a small number
  of particles.
• In reality, people often do novel activities or
  commit some errors
• Solution Use two trackers simultaneously and
  compute Bayes factors between the two models.

14
Experimental results

• 60 days of GPS data from one person using
  wearable GPS.
• First 30 days for learning and the rest for
  empirical comparison

15
Activity model learning
16
Infering Trip Segments
17
Empirical comparison to flat model
18
Comparison to 2MM model
19
Detection of user errors
20
Detection of user errors
21
Summary

• Paper introduces Hierarchical markov model that
  can learn and infer users daily movements.
• Model uses multiple levels of abstractions
  lowest level GPS, highest level transportation
  modes and goals.
• Rao-Blackwellised particle filtering used for
  inference
• Learning significant locations was done in an
  unsupervised manner using the EM algorithm.
• Novelty detection or abnormal behavior by model
  detection.