In cooperative video games, traditional AI companions are deployed to assist players, who control them using hotkeys or command wheels to issue predefined commands such as ''attack'', ''defend'', or ''retreat''. Despite their simplicity, these methods, which lack target specificity, limit players' ability to give complex tactical instructions and hinder immersive gameplay experiences. To address this, we propose the FPS AI Companion who Understands Language (F.A.C.U.L.), the first real-time AI system that enables players to communicate and collaborate with AI companions using natural language. By integrating natural language processing with a confidence-based framework, F.A.C.U.L. efficiently decomposes complex commands and interprets player intent. It also employs a dynamic entity retrieval method for environmental awareness, aligning human intentions with decision-making. Unlike traditional rule-based systems, our method supports real-time language interactions, enabling players to issue complex commands such as ''clear the second floor,'' ''take cover behind that tree,'' or ''retreat to the river''. The system provides real-time behavioral responses and vocal feedback, ensuring seamless tactical collaboration. Using the popular FPS game Arena Breakout: Infinite as a case study, we present comparisons demonstrating the efficacy of our approach and discuss the advantages and limitations of AI companions based on real-world user feedback.

With the rapid development of the gaming industry and the increasing popularity of multiplayer online games, realtime voice communication has become a crucial tool for team collaboration and tactical exchanges. Automatic Speech Recognition (ASR) technology plays a vital role in modern gaming by converting voice commands into text, enabling efficient communication among players. However, existing general-purpose ASR systems face significant challenges in gaming scenarios due to the unique characteristics of in-game communication, such as short phrases, rapid speech, game-specific jargon, and environmental noise. These limitations often lead to frequent recognition errors, increasing communication costs and negatively impacting the overall gaming experience. Furthermore, the scarcity of domainspecific ASR training data exacerbates these issues, hindering system optimization. To address the challenges of ASR systems in gaming scenarios, this study proposes the GO-AEC (Gaming-Oriented ASR Error Correction) framework. The framework leverages the generative capabilities of large language models (LLMs) and employs Retrieval-Augmented Generation (RAG) techniques with a gamespecific knowledge base to better adapt to gaming environments. Additionally, we introduce a data augmentation strategy that combines LLMs with text-to-speech (TTS) techniques to enhance the diversity and robustness of game-specific datasets. The GO-AEC framework consists of three key modules: the data augmentation module, the N-best hypothesis-based LLM correction module, and the dynamic knowledge base module powered by RAG. Experimental results demonstrate that, compared to baseline methods, the proposed framework reduces the character error rate (CER) by 6.22 % and the sentence error rate (SER) by 29.71 %. These findings indicate that the GO-AEC framework effectively addresses the challenges of ASR error correction in gaming scenarios.

In the realm of competitive gaming, 3D first-person shooter (FPS) games have gained immense popularity, prompting the development of game AI systems to enhance gameplay. However, deploying game AI in practical scenarios still poses challenges, particularly in large-scale and complex FPS games. In this paper, we focus on the practical deployment of game AI in the online multiplayer competitive 3D FPS game called Arena Breakout, developed by Tencent Games. We propose a novel gaming AI system named Private Military Company Agent (PMCA), which is interactable within a large game map and engages in combat with players while utilizing tactical advantages provided by the surrounding terrain. To address the challenges of navigation and combat in modern 3D FPS games, we introduce a method that combines navigation mesh (Navmesh) and shooting-rule with deep reinforcement learning (NSRL). The integration of Navmesh enhances the agent’s global navigation capabilities while shooting behavior is controlled using rule-based methods to ensure controllability. NSRL employs a DRL model to predict when to enable the navigation mesh, resulting in a diverse range of behaviors for the game AI. Customized rewards for human-like behaviors are also employed to align PMCA’s behavior with that of human players.

Deep Reinforcement Learning (DRL) agents have demonstrated impressive success in a wide range of game genres. However, existing research primarily focuses on optimizing DRL competence rather than addressing the challenge of prolonged player interaction. In this paper, we propose a practical DRL agent system for fighting games named _Shūkai_, which has been successfully deployed to Naruto Mobile, a popular fighting game with over 100 million registered users. _Shūkai_ quantifies the state to enhance generalizability, introducing Heterogeneous League Training (HELT) to achieve balanced competence, generalizability, and training efficiency. Furthermore, _Shūkai_ implements specific rewards to align the agent's behavior with human expectations. _Shūkai_'s ability to generalize is demonstrated by its consistent competence across all characters, even though it was trained on only 13% of them. Additionally, HELT exhibits a remarkable 22% improvement in sample efficiency. _Shūkai_ serves as a valuable training partner for players in Naruto Mobile, enabling them to enhance their abilities and skills.

Naruto Mobile is a popular mobile Fighting Game with over 100 million registered players. AI agents are deployed extensively to the game for a wide variety of applications such as level challenges and player training, which require them to fight like humans and imitate strong and weak players. Although deep reinforcement learning is an excellent approach to creating agents with diverse behaviors, it is difficult to apply to massive-scale games like Naruto Mobile which is built on a pool of more than 300 characters that have unique skills, speed, and attack range, as a traditional approach of self-play training at such scale may require a substantial computational cost and training time.In this paper, we present a new AI training approach called Heterogeneous Exploitation Self-Play (HESP) to improve AI agent generalization ability in Naruto Mobile and optimize its massive-scale self-play training so that the computational costs and train time are significantly reduced. The proposed algorithm has already been employed by the development team of Naruto Mobile to create AI agents, which, at the time of writing this paper, have been used in more than 300 million human-AI fighting matches. To the best of our knowledge, this is the first time that deep reinforcement learning has been employed by a commercial fighting game.

Dead reckoning (DR) is a key technique to increase scalability in Distributed Virtual Environments (DVE). Replacing data transmission with prediction, DR relies on its prediction capability to reduce the bandwidth consumption in the cost of inconsistency among participants. We propose a hot area targeting DR (HATDR) approach to increase the prediction capability by the hot area targeting pattern discovered with a noise-resistant clustering approach. This approach is shown to be robust against hyperparameters. Experiments carried out with a real-life MMOG dataset show that HATDR is comparable to the state-of-the-art DR approaches.

In recent years, Massively Multiplayer Online Games (MMOGs) are becoming popular, partially due to their sophisticated graphics and broad virtual world, and cloud gaming is demanded more than ever especially when entertaining with light and portable devices. This article considers the problem of server allocation for running MMOG on cloud, aiming to reduce the cost on cloud gaming service and meanwhile enhance the quality of service. The problem is formulated into minimizing an objective function involving the cost of server rental, the cost of data transfer and the network latency during the gaming time. A genetic algorithm is developed to solve the minimization problem for processing simultaneous server allocation for the players who log into the system at the same time while many existing players are playing the same game. Extensive experiments based on the player behavior in “World of Warcraft” are conducted to evaluate the proposed method and compare with the state-of-the-art as well. The experimental results show that the method gives a lower cost and a shorter network latency in most of the time.

We propose Distributed Neighbor Expansion (Distributed NE), a parallel and distributed graph partitioning method that can scale to trillion-edge graphs while providing high partitioning quality. Distributed NE is based on a new heuristic, called parallel expansion, where each partition is constructed in parallel by greedily expanding its edge set from a single vertex in such a way that the increase of the vertex cuts becomes local minimal. We theoretically prove that the proposed method has the upper bound in the partitioning quality. The empirical evaluation with various graphs shows that the proposed method produces higher-quality partitions than the state-of-the-art distributed graph partitioning algorithms. The performance evaluation shows that the space efficiency of the proposed method is an order-of-magnitude better than the existing algorithms, keeping its time efficiency comparable. As a result, Distributed NE can partition a trillion-edge graph using only 256 machines within 70 minutes.

Dead reckoning is an important technique used in distributed virtual environments (DVEs) to mitigate the bandwidth consumption of frequent state updates and the negative effects of network latency. This paper proposes a novel dead reckoning approach for common DVE applications such as multiplayer online games. Unlike traditional dead reckoning approaches that estimate the movements of remote entities with pure kinematic models, the new approach performs extrapolations with the considerations of environmental factors and human behaviours. We have performed experiments, based on a distributed car simulator, to compare the the new approach with representative existing dead reckoning approaches. The results show that the new approach gives more accurate predictions with an acceptable overhead.

Smart city projects, infrastructure planning, and traffic engineering are some of the applications where traffic simulations are playing an increasingly important role. Although many traffic simulators, commercial or open-sourced, are available at our disposal today, choosing the one that best fits a user's requirements is usually not possible by taking into account only the qualitative aspects and features of the simulator. In resource-constrained simulation platforms, performing traffic simulations with less memory usage and faster execution time is always highly coveted. In this paper, we propose a quantitative benchmarking approach for evaluating the performance of traffic simulator, based on commonplace scenarios and real-life city maps.

Continuous collision detection (CCD) is a process to interpolate the trajectory of polygons and detect collisions between successive time steps. However, primitive-level CCD is a very time-consuming process especially for a large number of moving polygons. Over the years, a number of approaches have been proposed to improve the computational efficiency of CCD by culling out the non-colliding primitives before exact overlap tests. These approaches have two fundamental disadvantages. First, they are mainly designed for self-and pairwise CCD and thus the performance gain would be limited when they are applied to large-scale scenes that contain thousands of moving polygons. Second, they are designed as sequential processes appropriate for execution on a single processor. Therefore, deploying them on high-performance parallel computing systems would not increase their computational efficiency.

The techniques of dead reckoning (DR) and interest management have been studied for more than two decades to optimize the data transmission on multiplayer online game networks. However, there is a lack of investigation in the integration of the two techniques, which may bring more benefits in terms of reducing bandwidth usage and enhancing computational efficiency. This paper presents a hybrid approach for efficient data distribution in multiplayer online games. We propose a new multi-threshold DR algorithm, which is more flexible than the traditional one-threshold DR approach. The new DR algorithm is combined with zone-based interest management (ZBIM) to take advantage of data filtering with minimum trade-offs. A number of experiments were conducted based on The Open Racing Car Simulator (TORCS) to evaluate the performance of the proposed hybrid approach. The results show that it performs better than the DR and ZBIM techniques in term of bandwidth usage without significant runtime difference.

Real-time geometric model distribution in large-scale distributed virtual environments (DVEs) is achieved by a technique known as content streaming. Content streaming mechanisms identify the fraction of static content to be delivered to each user, as DVE users would not be interested in the complete content of a virtual world at any point of time in the simulation. This technique also eliminates the hassles of pre-installations and offline patch updates. Most existing content streaming approaches reduce the overall network resource usage by compromising on the details of the 3D models streamed. In this paper, we present a caching framework termed as Local Data Management (LDM), which is used in conjunction with a multiresolution content filtering mechanism. The proposed framework reduces bandwidth usage by maintaining local copies of geometry data on the client system. We have evaluated the performance of the LDM framework in order to understand the trade-offs between resource utilization on the server and local disk space usage on the client, under various system conditions and scenarios.

This paper proposes a novel path-assisted dead reckoning algorithm for distributed virtual environments (DVEs). Unlike traditional dead reckoning algorithms that perform extrapolations based solely on kinematic models, the new algorithm takes environmental factors and human behaviours into account. Its design is based on an assumption that human-controlled entities tend to follow similar paths in the DVEs. Experimental evaluations, based on a car simulation, show that the path-assisted dead reckoning algorithm outperforms the traditional algorithm in terms of reducing the number of corrections with acceptable computational and space overhead.

Traffic simulation can be very computationally intensive, especially for microscopic simulations of large urban areas (tens of thousands of road segments, hundreds of thousands of agents) and when real-time or better than real-time simulation is required. For instance, running a couple of what-if scenarios for road management authorities/police during a road incident: time is a hard constraint and the size of the simulation is relatively high. Hence the need for distributed simulations and for optimal space partitioning algorithms, ensuring an even distribution of the load and minimal communication between computing nodes. In this paper we describe a distributed version of SUMO, a simulator of urban mobility, and SParTSim, a space partitioning algorithm guided by road network for distributed simulations. It outperforms classical uniform space partitioning in terms of road segment cuts and load-balancing.

As the scale of Distributed Virtual Environments (DVEs) grows in terms of participants and virtual entities, using interest management schemes to reduce bandwidth consumption becomes increasingly common for DVE development. The interest matching process is essential for most of the interest management schemes which determines what data should be sent to the participants as well as what data should be filtered. However, if the computational overhead of interest matching is too high, it would be unsuitable for real-time DVEs for which runtime performance is important. This paper presents a new approach of interest matching which divides the workload of matching process among a cluster of computers. Experimental evidence shows that our approach is an effective solution for the real-time applications.

Interest management is essential for real-time large-scale distributed virtual environments (DVEs) which seeks to filter irrelevant messages on the network. Many existing interest management schemes such as HLA DDM focus on providing precise message filtering mechanisms. However, this leads to a second problem: the computational overhead of the interest matching process. If the CPU cost of interest matching is too high, it would be unsuitable for real-time applications such as multiplayer online games for which runtime performance is important. This paper evaluates the performance of existing interest matching algorithms and proposes a new algorithm based on parallel processing. The new algorithm is expected to have better computational efficiency than existing algorithms and maintain the same accuracy of message filtering as them. Experimental evidence shows that our approach works well in practice.

Interest management is essential for scalable collaborative virtual environments (CVEs) which sought to reduce bandwidth consumption on the network. Most of the interest management systems such as Data Distribution Management (DDM) service of the High Level Architecture (HLA) concentrate on providing precise message filtering mechanisms. However, in doing so a second problem is introduced: the CPU cycle overheads of filtering process. If the cost in terms of computational resources of interest management itself is too high, it would be unsuitable for real time applications such as multiplayer online games (MOGs) for which runtime performance is important. In this paper we present a scalable interest management algorithm which is suitable for HLA DDM. Our approach employs the collision detection method of I-COLLIDE for fast interest matching. Furthermore, the algorithm has been implemented in our commercialized MOG middleware - Lucid Platform. Experimental evidence demonstrates that it works well in practice.