Constructivist Robots Assembling Value-Based Decision Heuristics

Conference Paper (2026)
Author(s)

Tyler Olson (TU Delft - Mechanical Engineering)

Alex Gabriel (TU Delft - Mechanical Engineering)

Carlos Hernández (TU Delft - Mechanical Engineering)

Research Group
Learning & Autonomous Control
More Info
expand_more
Publication Year
2026
Language
English
Research Group
Learning & Autonomous Control
Event
39th International Workshop on Qualitative Reasoning (2026-08-15 - 2026-08-15), Bremen, Germany
Page Views
15
Reuse Rights

Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.

Abstract

Modern robots make decisions in many ways, but rely on their designers to choose which strategies to employ and when. AI and robotic architectures commonly use a single decision-making strategy, usually based on either multiple regression, which
presumes the principle of total evidence, or optimization under constraints, which risks unbounded calculation of a stopping condition. When agents encounter situations violating these assumptions, their limited ability to adapt results in mission failure. Adopting a perspective of bounded rationality from the cognitive sciences, we propose a constructivist decision-making framework for autonomous agents to formulate their own decision strategies based on their mission-specific values. By semantically annotating different implementations of each modular component, it becomes possible to assemble and execute many different decision strategies. We represent each strategy as a behavior tree and assemble them using an automated theorem prover. For proof of concept, we simulate three example strategies used by a domestic robot performing a search and retrieval task, demonstrating that our method can assemble novel decision strategies and execute them on the order of a few seconds, and that it is applicable recursively to meta decisions.

Files

Paper_9.pdf
(pdf | 0.318 Mb)
License info not available