- Fuzzy Control For an Inverted Pendulum
Build a fuzzy control system to balance an inverted pendulum in a cart/pole system.
Given a simulation of an inverted pendulum on top of a cart that can move in 2
dimensions and provides information regarding position and velocity, design and
implement a fuzzy control system that can provide commands to the cart to successfully
balance the inverted pendulum and stay within a fixed region with the cart.
- Optimal Navigation Strategies with Uncertain Actions
Determine an approximately optimal navigation strategy for a
grid-based navigation task with probabilistic actions (left, right,
up, down, each with a 80% success probability) and uncertain state
observations (The location's x and y coordinate are returned with a
probability of 0.8 for the actual location and 0.1 for each of the
neighboring coordinates). Use the techniques introduced in the course
to track the belief state and to determine an approximately optimal
policy (in the presence of obstacles).
- Multi-Mean Estimation of Highly Uncertain Motion Data
Using the same setup as in Project 1, use a particle filter to
estimate the location and motion parameters of a cube where
observations consist of a noisy version of the distance of the cube
from the left upper corner of the box (the size of the box is provided).
- Estimation of Multiple Load Patterns
Two unknown load patterns (each with different expected packet arrival
rates) are present on the same network. Try to estimate both load
parameters and the occurrence of switching points between the
- Optimal Switching Under Varying Load Conditions
Two network connections of different bandwidth are available, each
with a different cost. Determine the optimal switching points given the
predicted amount of traffic produced by two randomly operating
processes. Here traffic conditions have to be predicted using a filter
and a switching policy has to be determined based on the predicted
load and the costs associated with the line and the delay caused by