Best-First Enumeration Based on Bounding Conflicts, and its Application to Large-scale Hybrid Estimation

Timmons, Eric; Williams, Brian

doi:10.1613/jair.1.11892

Cited by 4 publications

(10 citation statements)

References 26 publications

(35 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Furthermore, [13] models the hybrid system through a hybrid Bayesian network, and proposes a sampling-based approximation algorithm to track hybrid states. In parallel, [31], [12], [32], [33] model hybrid systems through a probabilistic hybrid automaton (PHA) [34] or a switching linear dynamical systems (SLDS) [35], and apply efficient pruning, search, and sampling methods to maintain reasonable estimation performance. Existing factored inference methods often assume a linear system to obtain closed-form or trackable solutions, and do not fully utilize the relevant context information.…”

Section: Related Workmentioning

confidence: 99%

“…2) Trajectory Sample Selection: We validate the effectiveness of our sample selection method, FPS, by comparing it with a few standard options, including i) a random sampler picking samples based on their weights, ii) a most-likely sampler that selects the top likely samples, similar to selecting the most-probable intent in [15] and best-k enumeration in [32], iii) a sampler based on non-maximum suppression (NMS), as used in [6], which selects samples greedily by finding the next sample that is distant enough from existing samples given a threshold. For a fair comparison, the distance measure in NMS is the same as FPS based on final locations, and we empirically choose 2 different distance thresholds (2 meters and 4 meters) to select the next sample.…”

Section: B Ablation Studymentioning

confidence: 99%

See 1 more Smart Citation

HYPER: Learned Hybrid Trajectory Prediction via Factored Inference and Adaptive Sampling

Huang¹,

Rosman²,

Gilitschenski³

et al. 2021

Preprint

View full text Add to dashboard Cite

Modeling multi-modal high-level intent is important for ensuring diversity in trajectory prediction. Existing approaches explore the discrete nature of human intent before predicting continuous trajectories, to improve accuracy and support explainability. However, these approaches often assume the intent to remain fixed over the prediction horizon, which is problematic in practice, especially over longer horizons. To overcome this limitation, we introduce HYPER, a general and expressive hybrid prediction framework that models evolving human intent. By modeling traffic agents as a hybrid discretecontinuous system, our approach is capable of predicting discrete intent changes over time. We learn the probabilistic hybrid model via a maximum likelihood estimation problem and leverage neural proposal distributions to sample adaptively from the exponentially growing discrete space. The overall approach affords a better trade-off between accuracy and coverage. We train and validate our model on the Argoverse dataset, and demonstrate its effectiveness through comprehensive ablation studies and comparisons with state-of-the-art models.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: B Ablation Studymentioning

confidence: 99%

HYPER: Learned Hybrid Trajectory Prediction via Factored Inference and Adaptive Sampling

Huang¹,

Rosman²,

Gilitschenski³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Our idea is to compute the likelihoods during the course of search over structures, and use what is learned from candidate structures to limit further evaluation to fewer structures. Our particular approach is based on the A*BC algorithm [136], which was introduced to incrementally improve a heuristic from evaluated candidates within hybrid estimation problems.…”

Section: Bounding Conflictsmentioning

confidence: 99%

“…We present a novel variant of A* with bounding conflicts [136] that substantially accelerates finding the optimal structural model. The algorithm uses a set of optimized likelihoods to place upper bounds on the likelihoods of other parent sets.…”

Section: Motivationmentioning

confidence: 99%

“…These rules are generalized to prune additional states from the search space. A* with bounding conflicts [136] further generalizes this idea, by deriving stricter bounds on the likelihoods of variable assignments as they are considered during search. These bounds are used to inform more precise heuristics, reducing the number of states that need to be considered searched.…”

Section: A* and Conflict-directed Searchmentioning

confidence: 99%

See 1 more Smart Citation

Query-driven adaptive sampling

Ayton¹

View full text Add to dashboard Cite

Automated information gathering allows exploration of environments where data is limited and gathering observations introduces risk, such as underwater and planetary exploration. Typically, exploration has been performed in service of a query, with a unique algorithm developed for each mission. Yet this approach does not allow scientists to respond to novel questions as they are raised. In this thesis, we develop a single approach for a broad range of adaptive sampling missions with risk and limited prior knowledge. To achieve this, we present contributions in planning adaptive missions in service of queries, and modeling multi-attribute environments. First, we define a query language suitable for specifying diverse goals in adaptive sampling. The language fully encompasses objectives from previous adaptive sampling approaches, and significantly extends the possible range of objectives. We prove that queries expressible in this language are not biased in a way that avoids information. We then describe a Monte Carlo tree search approach to plan for all queries in our language, using sample based objective estimators embedded within tree search. This approach outperforms methods that maximize information about all variables in hydrocarbon seep search and fire escape scenarios. Next, we show how to plan when the policy must bound risk as a function of reward. By solving approximating problems, we guarantee risk bounds on policies with large numbers of actions and continuous observations, ensuring that risks are only taken when justified by reward. Exploration is limited by the quality of the environment model, so we introduce Gaussian process models with directed acyclic structure to improve model accuracy under limited data. The addition of interpretable structure allows qualitative expert knowledge of the environment to be encoded through structure and parameter constraints. Since expert knowledge may be incomplete, we introduce efficient structure learning over structural models using A* search with bounding conflicts. By placing bounds on likelihood of substructures, we limit the number of structures that are trained, significantly accelerating search. Experiments modeling geographic data show that our model produces more accurate predictions than existing Gaussian process methods, and using bounds allows structure to be learned in 50% of the time.

show abstract

HYPER: Learned Hybrid Trajectory Prediction via Factored Inference and Adaptive Sampling

Huang¹,

Rosman²,

Gilitschenski³

et al. 2022

2022 International Conference on Robotics and Automation (ICRA)

View full text Add to dashboard Cite

Best-First Enumeration Based on Bounding Conflicts, and its Application to Large-scale Hybrid Estimation

Cited by 4 publications

References 26 publications

HYPER: Learned Hybrid Trajectory Prediction via Factored Inference and Adaptive Sampling

HYPER: Learned Hybrid Trajectory Prediction via Factored Inference and Adaptive Sampling

Query-driven adaptive sampling

HYPER: Learned Hybrid Trajectory Prediction via Factored Inference and Adaptive Sampling

Contact Info

Product

Resources

About