active perception planning

To address decision-making risks arising from uncertainty in surrounding agents’ behaviors within dynamic traffic environments, this paper proposes a risk-sensitive motion planning framework integrating multimodal trajectory prediction with active probing. Methodologically: (1) we design a novel analytically tractable risk metric, yielding closed-form solutions and finite-value guarantees under Gaussian mixture trajectory predictions; (2) we introduce an information-gain-driven active probing mechanism to dynamically reduce ambiguity in estimating other agents’ intent parameters; (3) we formulate a unified probabilistic planning objective that jointly optimizes trajectory generation and probing strategies. Evaluated on the large-scale MetaDrive simulator, our approach significantly improves navigation safety and decision robustness in complex interactive scenarios while maintaining compatibility with diverse traffic participant behavior models.

This work addresses the challenge of providing reachability guarantees in belief-space planning under motion uncertainty and state-control constraints. The authors propose PRISM, an algorithm that establishes, for the first time, a theory of constrained state covariance controllability, thereby ensuring full coverage in belief space while guaranteeing completeness within finite time and memory. PRISM decomposes planning into deterministic mean trajectory generation and covariance contraction, integrating multi-query belief roadmap construction with online local optimization. This approach achieves substantially improved performance: it attains 100% coverage in low- to moderate-difficulty scenarios and maintains 97–100% coverage even in the most challenging cases—significantly outperforming existing methods, all of which fall below 45%. Moreover, PRISM yields trajectories with lower cost and reduced variance.

This work addresses the challenge of robot planning under partial observability, where observation-dependent branching decisions render conventional sequential trajectories inadequate for handling uncertainty. The paper introduces tree-structured trajectories into partially observable model predictive control (MPC) and task and motion planning (TAMP), explicitly modeling multiple belief-state evolution paths induced by observations. Key contributions include a distributed augmented Lagrangian algorithm (D-AuLa) enabling parallel optimization, an extension of logical geometric programming (LGP) to support hierarchical decision-making in belief space, and a macro-action policy to enhance scalability. Experimental results demonstrate that the proposed approach significantly reduces control cost and meets real-time requirements in autonomous driving scenarios, validates effectiveness on small-scale problems, and extends to larger-scale applications through exploratory strategies.

This work addresses the integrated task and motion planning problem for robots with nonlinear dynamics operating in partially observable environments with semantic label uncertainty. It presents the first unified framework that combines LTLf-specification-guided task planning with kinodynamic motion planning under semantic uncertainty within a partially observable stochastic hybrid system formalism. The authors propose an anytime algorithm that synergistically integrates decision-theoretic reasoning with sampling-based motion planning, leveraging the system’s structural properties to enable efficient computation. The approach is theoretically guaranteed to be complete and asymptotically optimal. Experimental evaluations across diverse scenarios with varying degrees of semantic uncertainty demonstrate its superior performance over baseline methods, confirming both its effectiveness and robustness.

Traditional planning models goals as deterministic state subsets, failing to capture goal uncertainty arising from noisy perception, learning generalization, and other real-world uncertainties. Method: This paper proposes a distribution-to-distribution planning framework that directly represents both goals and current states as probability distributions, explicitly accommodating dynamic environmental uncertainty. It introduces goal distributions as fundamental planning primitives, integrates unscented transformation for nonlinear probabilistic kinematics modeling, and employs cross-entropy optimization to minimize the KL divergence between predicted and target distributions—enabling end-to-end uncertainty-aware planning. Contribution/Results: The framework unifies several classical goal cost functions as special cases. Extensive simulations demonstrate strong robustness against state disturbances, model mismatch, and data-driven goal uncertainty. Moreover, it significantly improves task success reliability under multimodal, sparse, and constrained goal distributions.

This work addresses a key limitation in existing active reconstruction methods, which often neglect motion uncertainty or prioritize tracking at the expense of reconstruction coverage when handling moving objects. To overcome this, the authors propose a novel next-best-view (NBV) planning framework that jointly accounts for both motion and measurement uncertainties. The approach uniquely integrates motion prediction into coverage-driven viewpoint selection by employing a fixed-lag Gaussian process smoother to estimate and probabilistically forecast the future state of the object from noisy observations. This predictive distribution is then used to evaluate the expected observation quality of candidate viewpoints, which are subsequently optimized under reachability constraints. Experimental results demonstrate that the proposed method significantly outperforms non-predictive NBV strategies and prediction-based approaches focused solely on tracking, achieving superior reconstruction completeness.

This work addresses safe navigation in dynamic environments with uncertain, time-varying obstacles by anticipating local observations. It introduces the first integration of precise contingency planning with Safe Interval Path Planning (SIPP) to generate formally verified safe macro-actions. The approach performs bounded AND/OR search over a cached action–observation graph to select optimal action sequences for each reachable observation. To guide search efficiently, it employs optimistic and robust SIPP relaxations that yield admissible heuristic bounds. Decisions are made dynamically based on local observations, enabling real-time adaptation. Experiments demonstrate superior performance over fixed-path baselines in controlled road networks and successful planning in gated scenarios where conservative methods fail. The study also reveals a scalability bottleneck as observation uncertainty increases.

This work addresses motion planning for continuous-time stochastic systems under both process and observation uncertainties by proposing a sampling-based planning framework that enables continuous-time probabilistic safety verification over entire trajectories. The approach constructs an offline hybrid belief propagation model that integrates continuous-time ordinary differential equation (ODE) dynamics with discrete Kalman updates, and introduces a belief barrier function as a safety checker capable of detecting potential constraint violations between sampling instants—marking the first method to achieve such intra-interval safety guarantees. Integrated with RRT/SST planners, the framework demonstrates superior performance over conventional discrete-time methods across multiple benchmark scenarios, including narrow passages, achieving higher success rates, enhanced robustness, and stronger formal safety assurances.

This work addresses the optimal observability problem (OOP) in uncertain environments, which entails balancing task feasibility against sensing costs. Focusing on its decidable subproblems—sensor selection (SSP) and position observability (POP)—the paper proposes a novel solution framework based on POMDP decomposition, integrating parameter synthesis with a symbolic–subsymbolic hybrid approach. This method dramatically improves computational efficiency, scaling solvable instances by three orders of magnitude and reducing runtime by five orders of magnitude compared to prior techniques. Consequently, the approach substantially expands the tractable boundary of observability-aware planning in partially observable settings.

This work addresses the limitation of existing motion planning methods that often neglect terminal state quality and struggle to reliably achieve goals under uncertainty. The authors propose a unified motion planning framework that explicitly integrates terminal cost with cumulative trajectory cost, introducing terminal cost for the first time into asymptotically optimal kinodynamic planning (AO-RRT) while preserving its theoretical optimality guarantees. The approach is further extended to belief space by formulating goal achievement as the optimization of a lower bound on success probability, derived using the Wasserstein distance. Leveraging data-driven models of dynamics and process uncertainty, the method demonstrates significantly improved goal achievement rates in both simulation and real-world experiments across diverse tasks—including Flappy Bird, autonomous parking, and planar pushing—under uncertain conditions.