arXiv subjects

Pranay Meshram

Publications and source records attributed to Pranay Meshram.

4 recordsLinked to original sources

arXiv2026

COMBAT: Conditional World Models for Behavioral Agent Training

Recent advances in video generation have spurred the development of world models capable of simulating 3D-consistent environments and interactions with static objects. However, a significant limitation remains in their ability to model dynamic, reactive agents that can intelligently influence and interact with the world. To address this gap, we introduce COMBAT, a real-time, action-controlled world model trained on the complex 1v1 fighting game Tekken 3. Our work demonstrates that diffusion models can successfully simulate a dynamic opponent that reacts to player actions, learning its behavior implicitly. Our approach utilizes a 1.2 billion parameter Diffusion Transformer, conditioned on latent representations from a deep compression autoencoder. We employ state-of-the-art techniques, including causal distillation and diffusion forcing, to achieve real-time inference. Crucially, we observe the emergence of sophisticated agent behavior by training the model solely on single-player inputs, without any explicit supervision for the opponent's policy. Unlike traditional imitation learning methods, which require complete action labels, COMBAT learns effectively from partially observed data to generate responsive behaviors for a controllable Player 1. We present an extensive study and introduce novel evaluation methods to benchmark this emergent agent behavior, establishing a strong foundation for training interactive agents within diffusion-based world models.

cs.CV↗

arXiv2026

CLEAR: A Semantic-Geometric Terrain Abstraction for Large-Scale Unstructured Environments

Long-horizon navigation in unstructured environments demands terrain abstractions that scale to tens of square kilometers while preserving semantic and geometric structure. Grids scale poorly, and quadtrees misalign with terrain boundaries. Neither encodes terrain semantics essential for traversability-aware planning, yielding infeasible or inefficient paths for autonomous ground vehicles operating over more than 10 square kilometers. CLEAR (Connected Landcover Elevation Abstract Representation) is a reusable terrain abstraction framework for large-scale planning that produces convex, semantically aligned regions encoded as a terrain-aware graph. Evaluated on digital terrain maps spanning 9 to 100 square kilometers with physics-based simulation, CLEAR reduces per-query planning by 2x to 19.6x over unabstracted raw-grid A-star after one-time abstraction, with 6.7 percent cost overhead. In physics-based simulation, CLEAR delivers 5 to 8.6 percent shorter executed paths and 100 percent task completion across all maps, compared with 90 to 100 percent for Quadtree. These results hold against AMRA-star, a strong anytime multi-resolution planner, and generalize to a learned elevation-driven cost, demonstrating CLEAR's scalability and cost-model independence as a reusable planning layer.

cs.RO↗

arXiv2025

QAL: A Loss for Recall Precision Balance in 3D Reconstruction

Volumetric learning underpins many 3D vision tasks such as completion, reconstruction, and mesh generation, yet training objectives still rely on Chamfer Distance (CD) or Earth Mover's Distance (EMD), which fail to balance recall and precision. We propose Quality-Aware Loss (QAL), a drop-in replacement for CD/EMD that combines a coverage-weighted nearest-neighbor term with an uncovered-ground-truth attraction term, explicitly decoupling recall and precision into tunable components. Across diverse pipelines, QAL achieves consistent coverage gains, improving by an average of +4.3 pts over CD and +2.8 pts over the best alternatives. Though modest in percentage, these improvements reliably recover thin structures and under-represented regions that CD/EMD overlook. Extensive ablations confirm stable performance across hyperparameters and across output resolutions, while full retraining on PCN and ShapeNet demonstrates generalization across datasets and backbones. Moreover, QAL-trained completions yield higher grasp scores under GraspNet evaluation, showing that improved coverage translates directly into more reliable robotic manipulation. QAL thus offers a principled, interpretable, and practical objective for robust 3D vision and safety-critical robotics pipelines

cs.CV↗

arXiv2023

Empir3D : A Framework for Multi-Dimensional Point Cloud Assessment

Advancements in sensors, algorithms, and compute hardware have made 3D perception feasible in real time. Current methods to compare and evaluate the quality of a 3D model, such as Chamfer, Hausdorff, and Earth-Mover's distance, are uni-dimensional and have limitations, including an inability to capture coverage, local variations in density and error, and sensitivity to outliers. In this paper, we propose an evaluation framework for point clouds (Empir3D) that consists of four metrics: resolution to quantify the ability to distinguish between individual parts in the point cloud, accuracy to measure registration error, coverage to evaluate the portion of missing data, and artifact score to characterize the presence of artifacts. Through detailed analysis, we demonstrate the complementary nature of each of these dimensions and the improvements they provide compared to the aforementioned uni-dimensional measures. Furthermore, we illustrate the utility of Empir3D by comparing our metrics with uni-dimensional metrics for two 3D perception applications (SLAM and point cloud completion). We believe that Empir3D advances our ability to reason about point clouds and helps better debug 3D perception applications by providing a richer evaluation of their performance. Our implementation of Empir3D, custom real-world datasets, evaluations on learning methods, and detailed documentation on how to integrate the pipeline will be made available upon publication.

cs.CV↗