Closed rayvburn closed 1 year ago
Robot trajectory has to be investigated individually in cost functions, prediction in the planner does not make sense.
Possible approaches for people/dynamic objects prediction:
struct PersonWithTrajPrediction
- predict once the new data is receivedWorld
class needs to be completely redesigned to make it right throughout the modules (e.g., SFM). To make it working properly, World
will be slightly modified as a "quick fix".
For now, let's leave that version. Later, obstacles, people and groups must be explicitly incorporated into World
model instead of static and dynamic objects.
Related to
HeadingDisturbanceCostFunction
andPersonalSpaceIntrusionCostFunction
(and possiblyFformationSpaceIntrusionCostFunction
in the future).Aims to optimize usage of resources.