Download - Databases for Robotics Applications
Databases for Robotics Applications
Thomas Young
Presentation Outline
Introduction Database Solutions Spatio Temporal Databases TinyDB
Introduction
Various types of Database Applications Bipedal Robot Research Obstacle Database Sensor Networks Moving Objects Database
Bipedal Robot Research
Training a bipedal robot to walk
Bipedal Robot Research
Bipedal Robot Research
Components Servo Motors Sensors PD Motor Control Neural Net Database
Bipedal Robot Research
Learning Process
Obstacle Database
Database of natural obstacles CFIT Problem in Aviation Terrain Awareness System
Obstacle Database
Obstacle Database
eTAWS Database of man made obstacles including bridges, towers,
overpasses, hydro lines, buildings
Obstacle Database
Standard relational database Spatial representations stored as vectors or rasters
using an extended spatial type Uses SQL queries
Sensor Networks
Networks of either homogeneous or heterogenous sensor types
Sensors characterized by power, computation, communications
Networks characterized by configuration, types of sensors
Harvard Motes Smart Dust
Sensor Networks
Homogeneous Sensor Network
Moving Objects Database
Database of objects that change position in time and space
Tracking of vehicles, assets, people, animals Fleet tracking Scientific research Surveillance
Moving Objects Database
Vehicular Traffic
Moving Objects Database
Firefighting Assets
Moving Objects Database
Ground Forces
Database Solutions
NOSQL TinyDB Spatio Temporal Databases
NOSQL
No ACID guarantee Distributed fault tolerant architecture Do not follow a fixed schema Performance and scalability
TinyDB
Sensor Networks with nodes running TinyOS Runs TinySQL (subset of SQL) Extensible framework for attributes, commands, and
aggregates Interacts with sensor network as a whole Multiple concurrent queries Entire sensor network is infinitely long table Tuples consist of individual sensor and attributes
TinyDB
Energy Cost of a query that selects 100 tuples is less than the cost of a single packet transmission
Spatio Temporal Databases
Objects that move in space and time Handle queries that index by an object, time or time interval,
physical location
(a) (b) a moving point a moving and shrinking region
y
t
x
y
t
x
Spatio Temporal Databases
Query Examples Find all objects in a given area at a given time Find all objects in a given area between these times Find which object was closest to position X at time T How many objects passed through area A at time T Given spatio-temporal relationships R1 and R2, find
out which pairs intersected between T1 and T2
Spatio Temporal Databases
R Trees
Spatio Temporal Databases
Historical R Tree
x
t t1 t3 t2
y
o2
o3
Query region Q
o1
t4 tnow
Spatio Temporal Databases
How to store “now”? Use a large value… Long lived objects will have very long MBRs,
difficult to cluster Extensive overlap and empty space bad
query performance for specific queries Use partiallly persistent R-tree Multi-version Binary Tree applied to R-tree
Spatio Temporal Databases
Trees at consecutive timestamps may share branches to save space
Spatio Temporal Databases
Trees at consecutive timestamps may share branches to save space.
Spatio Temporal Databases
HR-trees answer timestamp queries very efficiently. A timestamp query degenerates into a spatial window query handled by
the corresponding R-tree at the query timestamp.
Not quite efficient: Expensive space consumption. A node needs to be duplicated even when only one object moves. Interval query processing is inefficient. Although redundancy (from duplication) is necessary to maintain good
timestamp query performance, it is excessive in HR-trees
Spatio Temporal Databases
What if you want to track only one object? Use artificial deletes to get rid of others Approximate the object using many small MBRs This uses more space Instead split the areas into minimum number of
MBRs that contain the objects that move the most If object has constant velocity then equidistant splits Given x splits the best splits can be determined in
O(xlogn) time
THE END
THANK YOU
References
TinyDB Design Code and Implementations, Prakash Achutaramaiah Implementation and Research Issues in Query Processing for Wireless Sensor Networks,
Wei Hong, Sam Madden An On-Line Biped Mini-Robot Motion Learning Using Neural Network and Database
Management, Shih Fen Cheng, 2011 Seventh International Conference on Natural Computing
Towards Sensor Database Systems, Bonnet Phillipe, Gehrke Johannes, Seshadri Praveen Distributed Sensor Databases for Multi-robot Teams, Cowley Anthony, Hwa-Chow Hsu,
Camillo J, Taylor Future Robotics Database Management System Alonw With Cloud TPS, Vijaykumar S,
Sarvanakumar S G, International Journal on Cloud Computing: Services and Architecture (IJCCSA), Vol 1, No.3, Novermber 2011
[Tao & Papadias 01]:MV3R-Tree: A Spatio-Temporal Access Method for Timestamp and Interval Queries. VLDB 2001: 431-440