Recent breakthroughs in Large Language Models (LLMs) have positioned them as a promising paradigm for agents, with long-term planning and decision-making emerging as core general-purpose capabilities for adapting to diverse scenarios and tasks. Real-time strategy (RTS) games serve as an ideal testbed for evaluating these two capabilities, as their inherent gameplay requires both macro-level strategic planning and micro-level tactical adaptation and action execution. Existing RTS game-based environments either suffer from relatively high computational demands or lack support for textual observations, which has constrained the use of RTS games for LLM evaluation. Motivated by this, we present TowerMind, a novel environment grounded in the tower defense (TD) subgenre of RTS games. TowerMind preserves the key evaluation strengths of RTS games for assessing LLMs, while featuring low computational demands and a multimodal observation space, including pixel-based, textual, and structured game-state representations. In addition, TowerMind supports the evaluation of model hallucination and provides a high degree of customizability. We design five benchmark levels to evaluate several widely used LLMs under different multimodal input settings. The results reveal a clear performance gap between LLMs and human experts across both capability and hallucination dimensions. The experiments further highlight key limitations in LLM behavior, such as inadequate planning validation, a lack of multifinality in decision-making, and inefficient action use. We also evaluate two classic reinforcement learning algorithms: Ape-X DQN and PPO. By offering a lightweight and multimodal design, TowerMind complements the existing RTS game-based environment landscape and introduces a new benchmark for the AI agent field.

Dawei Wang, Di Zhao, Xinyuan Liu, Marci Chi Ma, Xiaoyang Liu, Chengming Zhou, Gary Ushaw, Richard Davison. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Summary Advancement of computing and communication techniques transforms the traditional transport system into the intelligent transportation system (ITS). The development of distributed computing in a vehicular network platform also called Vehicular Edge Computing (VEC) promise to address most of the challenges faced by the ITS. Localization is important in these vehicular networks because of its key contribution in autonomous driving, smart traffic monitoring, and collision avoidance services. For localization, current GPS and hybrid methods are in‐efficient because of GPS outage in urban infrastructure and dynamic nature of the vehicular networks. The cooperative localization approaches, on the other hand, use dedicated short range communication to broadcast messages and estimate location. However, these messages are un‐encrypted and periodic which gives a privacy risk for vehicles. This article presents a privacy‐preserving cooperative localization in vehicular network based upon dynamic pseudonym changing strategy. First, the localization delay is addressed with the implementation of dynamic vehicular edge assignment for computational task management. In the next step, the localization is estimated from the neighbor and road side unit ranging measurement followed by a real‐time prediction of the vehicle. The performance of the proposed algorithms is analyzed in terms of localization accuracy and privacy preservation strength. Furthermore, the proposed method is simulated in a real city scenario followed by localization accuracy and privacy analysis. Finally, the localization accuracy and privacy strength of the proposed approach are compared with the state‐of‐the‐art methods.