Which campus is CS at KCL

WS 05/06:
Artificial intelligence lecture







Lecturer: PD Dr. Stefan Edelkamp Supervisor: M.Sc. Shahid Jabbar Times: Wed 14-16, Fri 12-14 (South Campus, HG I / HS 3) Exercises: preliminary Fri 16-18 (GB IV, R 126)


The lecture gives a problem- and algorithm-oriented introduction to methods of artificial intelligence (AI). Important sub-areas of modern AI research are addressed: problem solving through search, logic-based agents, action planning, two- and multi-person games, acting under uncertainty, machine learning, and robotics. The aim is to give students an overview of the various AI techniques.


It starts with a general introduction to AI by addressing various current AI projects. The main part of the lecture deals with problem-solving strategies and presents various search strategies. Action planning is introduced as a search in the state space. Probability theory helps to construct agents that can work with uncertain knowledge. The lecture deals with methods of machine learning and applications in robotics.
  1. As an introduction: attempts to define AI; Presentation of various current AI projects; Introduction to (intelligent) agents: reflex agents, goal-based agents, benefit-based agents and their environments.
  2. For problem solving: problem presentation; Simple search strategies: breadth-first search, cost-optimized breadth-first search, depth-first search, depth-first search with maximum depth, progressive deepening, bidirectional search; Informed search: best search, greedy search, A *, IDA * search; Heuristics; Improved algorithms: mountaineering, simulated annealing.
  3. On logic: representation and conclusions; Propositional logic: syntax, semantics, inference mechanisms; First level logic: syntax, semantics, inference mechanisms, situation calculus, frame problem, resolution proof, normal forms, resolution strategies, completeness.
  4. For planning: planning in the calculus of the situation; Forms of representation; Partial order planning; Graph planning; Feasibility plans; BDD planning; and planning through heuristic search.
  5. On probability theory: Insecure knowledge; Design of a decision theory agent; Baye's rules; Probability-based inference systems; Belief networks; and inference on belief networks.
  6. On machine learning: decision trees; Neural and Bayesian Networks; Learning of macro operators and control rules.
  7. On robotics: examples from industrial robots to humanoid robots; Types of sensor information: sonar, laser, GPS, camara, odometry; Self-localization, integration of information through Kalman filters; Map creation; Path planning


The slides of the lecture are relevant for the examination. The book template serves to deepen the lecture
  • Stefan Edelkamp, ​​Stefan Schroedl, and Sven Koenig
    Theory and Practice of Heuristic Search
    Submitted to Morgan Kaufmann