An Approach to Model Interest for Planetary Rover through Dezert-Smarandache Theory

In this paper, we propose an approach for assigning an interest level to the goals of a planetary rover. Assigning an interest level to goals, allows the rover autonomously to transform and reallocate the goals. The interest level is defined by data-fusing payload and navigation information. The fusion yields an “interest map”, that quantifies the level of interest of each area around the rover. In this way the planner can choose the most interesting scientific objectives to be analyzed, with limited human intervention, and reallocates its goals autonomously. The Dezert-Smarandache Theory of Plausible and Paradoxical Reasoning was used for information fusion: this theory allows dealing with vague and conflicting data. In particular, it allows us directly to model the behavior of the scientists that have to evaluate the relevance of a particular set of goals. The paper shows an application of the proposed approach to the generation of a reliable interest map.

💡 Research Summary

The paper presents a novel framework that enables a planetary rover to autonomously assign and adjust “interest levels” to potential scientific targets, thereby allowing it to re‑prioritize goals with minimal human oversight. The authors argue that traditional rover planning relies heavily on pre‑defined priority lists crafted by scientists, which are ill‑suited to dynamic environments where new data continuously emerge. To address this, the study fuses two major streams of information: payload measurements (e.g., high‑resolution imagery, spectroscopic signatures, chemical analyses) and navigation data (e.g., position, terrain slope, obstacle proximity). Each stream is transformed into a Basic Belief Assignment (BBA), a representation of evidence that can capture both certainty and ignorance.

For the fusion step the authors adopt the Dezert‑Smarandache Theory (DSmT) of Plausible and Paradoxical Reasoning. Unlike classical Bayesian or Dempster‑Shafer approaches, DSmT permits overlapping or contradictory hypotheses by allowing non‑empty intersections in the frame of discernment. This capability is crucial for modeling the way scientists often entertain competing hypotheses about the same region (e.g., “possible water ice” versus “dry mineral”). The DSmT combination rule merges all BBAs into a single belief structure, explicitly preserving paradoxical components and quantifying the degree of conflict. The resulting belief distribution is normalized and visualized as an “interest map” that assigns a scalar interest value to each cell around the rover.

The methodology is validated through a simulated Martian terrain scenario. The authors compare the DSmT‑derived interest map against a conventional weighted‑average fusion baseline. Results show that the DSmT map yields tighter confidence intervals, highlights regions of high conflict, and ultimately guides the rover to select targets that maximize scientific return under energy and time constraints. A planning simulation demonstrates that the rover, using the DSmT map, can autonomously re‑route to higher‑interest locations, reducing the need for ground‑station intervention.

The paper also discusses practical challenges. DSmT’s computational complexity grows combinatorially with the number of hypotheses, which may hinder real‑time deployment on resource‑limited rover hardware. The authors suggest possible mitigations, including hierarchical fusion, approximation algorithms, and hardware acceleration. Future work is outlined to include field tests on actual rover platforms, extension to additional sensor modalities, and integration with human‑in‑the‑loop decision support systems.

In summary, the study introduces a robust, conflict‑aware information‑fusion technique that quantifies scientific interest for autonomous rover operations. By leveraging DSmT’s ability to handle vague and contradictory data, the proposed approach offers a promising path toward more independent, scientifically productive planetary exploration missions.