MCHA4400
10 units
4000 level
Course handbook
Description
Vision-based sensors such as lidar, radar, and optical & thermal cameras are having an enormous impact on navigation of autonomous vehicles and mobile robotics. Navigation relies on accurate sensing models in combination with a 3D representation of the environment and a dynamic model of the vehicle and obstacles. Students who complete this course will acquire background geometric tools and kinematics for developing sensor likelihood models, which will be used within a Bayesian data fusion framework. This delivers estimates of the vehicle pose, pose-rate, and provides a map of the environment and includes topics such as simultaneous localisation and mapping (SLAM) and optic-flow egomotion (visual odometry).
Availability2024 Course Timetables
Callaghan
- Semester 2 - 2024
Learning outcomes
On successful completion of the course students will be able to:
1. Define appropriate reference frames and coordinate systems to express world-fixed and body-fixed objects and their motion
2. Calibrate sensor likelihood functions based on calibration data by posing and solving a constrained optimisation problem
3. Design a landmark-based SLAM solution
4. Implement and validate a real-time landmark-based SLAM solution
5. Compute optic flow on the view sphere based on a sequence of images
6. Design an optic-flow based navigation solution
7. Implement and validate a realtime optic-flow based navigation solution
Content
Fundamentals of visual sensors
- Revision of kinematics
- Geometry of vision
- Camera models (parametric and non-parametric)
- Planar and spherical projections
- Sensor calibration
Vision as pose sensor
- Feature identification / extraction
- Landmark management
- Data association
- Bundle adjustment
- Review of nonlinear Bayesian filtering
- Sparse extended information filter SLAM (SEIF-SLAM)
Vision as pose-rate sensor
- Image flow (sparse and dense estimators)
- Optic flow on the view sphere
- Egomotion
- Flow-based navigation
Assumed knowledge
MCHA4100 Mechatronics Systems, ENGG3300 Machine Learning for Engineers
Assessment items
Tutorial / Laboratory Exercises: Laboratory Exercise (x6)
Written Assignment: Visual SLAM
Written Assignment: Optic Flow Integration
Contact hours
Semester 2 - 2024 - Callaghan
Laboratory-1
- Face to Face On Campus 2 hour(s) per week(s) for 13 week(s) starting in week 1
Laboratory-2
- Face to Face On Campus 2 hour(s) per week(s) for 13 week(s) starting in week 1
Lecture-1
- Face to Face On Campus 2 hour(s) per week(s) for 13 week(s) starting in week 1
Course outline
Course outline not yet available.
The University of Newcastle acknowledges the traditional custodians of the lands within our footprint areas: Awabakal, Darkinjung, Biripai, Worimi, Wonnarua, and Eora Nations. We also pay respect to the wisdom of our Elders past and present.