@adrian please fix visuals.. ,)

minor bug fixes in node connecting algorithm

minor bug fixes in node connecting algorithm

All starting positions are valid.
Check for feasibility of map is not yet done.

introduced cities and intersections. These will make up the sparse network. THen you can add trainstations to the cities and populate them with tasks for agents (start/target). Orientation of agents needs to be fixed. Also check for invalid transitions in cities and nodes needs to be implemented.

Merge branch '147_new_level_generator' of gitlab.aicrowd.com:flatland/flatland into 147_new_level_generator

# Conflicts:
#	tests/test_flatland_env_sparse_rail_generator.py

introduced cities and intersections. These will make up the sparse network. THen you can add trainstations to the cities and populate them with tasks for agents (start/target). Orientation of agents needs to be fixed. Also check for invalid transitions in cities and nodes needs to be implemented.

not functional yet

updated malfunction behavior. Now a train can break at any point during its travel. We don't count number of stops anymore. This is to simulate more complex stochastic events.

# Conflicts:
#   examples/training_example.py

143 multi speed initialization

See merge request flatland/flatland!144

140 predictor multi speed

See merge request !141

# Conflicts:
#   examples/training_example.py