Enabling robots to communicate their objectives
The DOI matches its Crossref record; 7 of the 7 references it cites were found in a registry or an index.
Abstract
The overarching goal of this work is to efficiently enable end-users to correctly anticipate a robot’s behavior in novel situations. And since a robot’s behavior is often a direct result of its underlying objective function, our insight is that end-users need to have an accurate mental model of this objective function in order to understand and predict what the robot will do. While people naturally develop such a mental model over time through observing the robot act, this familiarization process may be lengthy. Our approach reduces this time by having the robot model how people infer objectives from observed behavior, in order to then show those behaviors that are maximally informative. We introduce two factors to define candidate models of human inference, and show that certain models indeed produce example robot behaviors that better enable users to anticipate what it will do in novel situations. Our results also reveal that choosing the appropriate model is key, and suggest that our candidate models do not fully capture how humans extrapolate from examples of robot behavior. We leverage these findings to propose a stronger model of human learning in this setting, and conclude by analyzing the impact of different ways in which the assumed model of human learning may be incorrect.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
An analysis of approximations for maximizing submodular set functions—I
Nemhauser, G. L., Wolsey, L. A., & Fisher, M. L. (1978) · Mathematical Programming · doi:10.1007/bf01588971 · Cited by 4,412
Context theory of classification learning
Medin, D. L., & Schaffer, M. M. (1978) · Psychological Review · doi:10.1037/0033-295x.85.3.207 · Cited by 3,061 · cited 1×
“This is motivated by the observation that humans seem to use such a nearest-neighbor, or exemplar-based, approach in other domains, in particular category learning (Medin and Schaffer 1978; Nosofsky 1986).”
Joint action: bodies and minds moving together
SEBANZ, N., BEKKERING, H., & KNOBLICH, G. (2006) · Trends in Cognitive Sciences · doi:10.1016/j.tics.2005.12.009 · Cited by 1,950
(2000) · doi:10.2460/ajvr.67.2.323 · Cited by 1,621
Action understanding as inverse planning
Baker, C. L., Saxe, R., & Tenenbaum, J. B. (2009) · Cognition · doi:10.1016/j.cognition.2009.07.005 · Cited by 920
Vul (2014) · doi:10.1111/cogs.12101 · Cited by 502
New Potentials for Data‐Driven Intelligent Tutoring System Development and Optimization
Koedinger, K. R., Brunskill, E., Baker, R. S. J. D., McLaughlin, E. A., & Stamper, J. (2013) · AI Magazine · doi:10.1609/aimag.v34i3.2484 · Cited by 178 · cited 2×
“In order to choose the most informative example behaviors for communicating a robot’s objective function to humans, we take an algorithmic teaching approach [3, 10, 14, 22, 32, 34]: we model how humans make inferences about the robot’s objective function from examples of its optimal behavior, and use this model to gen…”
Method
Cited by
Kocielnik, R., Amershi, S., & Bennett, P. N. (2019) · Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3290605.3300641 · Cited by 277
Towards Human-Centered Explainable AI: A Survey of User Studies for Model Explanations
Rong, Y., Leemann, T., Nguyen, T.-T., Fiedler, L., Qian, P., Unhelkar, V., Seidel, T., Kasneci, G., & Kasneci, E. (2024) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2023.3331846 · Cited by 278
Explainable AI and Reinforcement Learning—A Systematic Review of Current Approaches and Trends
Wells, L., & Bednarz, T. (2021) · Frontiers in Artificial Intelligence · doi:10.3389/frai.2021.550030 · Cited by 185
“…with one paper using 3 experts (Wang et al., 2018), others with students (Iyer et al., 2018), n = 40; and Greydanus et al. (2018), n = 31, and three recruiting using Amazon Mechanical Turk 3 (Huang et al., 2019, n = 191; Madumal et al., 2020, n = 120; and Ehsan et al., 2019, n = 65 and n = 60).”
BackgroundMethodLocal and global explanations of agent behavior: Integrating strategy summaries with saliency maps
Huber, T., Weitz, K., André, E., & Amir, O. (2021) · Artificial Intelligence · doi:10.1016/j.artint.2021.103571 · Cited by 53
“Some methods choose states that best enable the reconstruction of the original policy, using computational models such as inverse reinforcement learning (inferring the agent’s reward function) or imitation learning (constructing a mapping from states to agents’ actions) [35, 43].”
MethodHannibal, G., Weiss, A., & Charisi, V. (2021) · 2021 30th IEEE International Conference on Robot & Human Interactive Communication (RO-MAN) · doi:10.1109/ro-man50785.2021.9515513 · Cited by 11
Metadata
- Authors
- Huang, S. H., Held, D., Abbeel, P., & Dragan, A. D.
- Venue
- Autonomous Robots
- Published
- 2018-06-18
- Volume
- 43
- Issue
- 2
- Pages
- 309-326
- ISSN
- 0929-5593 (ISSN-L)
- 1573-7527
- OpenAlex
- W2594227402
- Cited by
- 101
- References
- 7
- FWCI
- 13.53
- Type
- journal article
- Language
- en
- Open access
- closed
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.