PACER: Preference-conditioned All-terrain Costmap Generation
This addresses the limitation of semantics-based approaches in handling new terrains for robot navigation, offering a more flexible solution, though it is incremental as it builds on machine-learning alternatives.
The paper tackles the problem of adapting terrain cost assignment in autonomous robot navigation to new user preferences without retraining, by introducing PACER, which generates costmaps from a single BEV image and preference context, showing it adapts quickly and generalizes better to novel terrains than existing methods.
In autonomous robot navigation, terrain cost assignment is typically performed using a semantics-based paradigm in which terrain is first labeled using a pre-trained semantic classifier and costs are then assigned according to a user-defined mapping between label and cost. While this approach is rapidly adaptable to changing user preferences, only preferences over the types of terrain that are already known by the semantic classifier can be expressed. In this paper, we hypothesize that a machine-learning-based alternative to the semantics-based paradigm above will allow for rapid cost assignment adaptation to preferences expressed over new terrains at deployment time without the need for additional training. To investigate this hypothesis, we introduce and study PACER, a novel approach to costmap generation that accepts as input a single birds-eye view (BEV) image of the surrounding area along with a user-specified preference context and generates a corresponding BEV costmap that aligns with the preference context. Using both real and synthetic data along with a combination of proposed training tasks, we find that PACER is able to adapt quickly to new user preferences while also exhibiting better generalization to novel terrains compared to both semantics-based and representation-learning approaches.