Paper: "Learning to Interpret Natural Language Navigation Instructions from Observations,"
David L. Chen and Raymond J. Mooney.
In Proceedings of the Twenty-Fifth AAAI Conference on Artificial Intelligence (AAAI-2011), pages 859--865, 2011.

Reason for choice: Basically, this one seemed interesting to me. It is related to navigation and deals with natural language.

Abstract: The ability to understand natural-language instructions is critical to building intelligent agents that interact with humans. We present a system that learns to transform natural-language navigation instructions into executable formal plans. Given no prior linguistic knowledge, the system learns by simply observing how humans follow navigation instructions. The system is evaluated in three complex virtual indoor environments with numerous objects and landmarks. A previously collected realistic corpus of complex English navigation instructions for these environments is used for training and testing data. By using a learned lexicon to refine inferred plans and a supervised learner to induce a semantic parser, the system is able to automatically learnto correctly interpret a reasonable fraction of the complex instructions in this corpus.

URL: Download the paper at this location
Alternate site: pdf

PRESENTATION:

Paper: "Online Graph Pruning for Pathfinding on Grid Maps,"
Daniel Harabor and Alban Grastien.
In Proceedings of the Twenty-Fifth AAAI Conference on Artificial Intelligence (AAAI-2011), pages 1114-1119, 2011.

Reason for choice: Looking through the papers, it was clear I didn't have the background to understand many of them from their titles alone. However, this one sounded more promising. I understand pathfinding fairly well, having implemented some of my own pathfinding algorithms. In fact, I've implemented algorithms specifically to solve grid-based map pathfinding which is what this paper is about. From reading the abstract, the paper claims that their approach can speed up A* by "an order of magnitude." That's a pretty bold claim, so I'm looking forward to reading the paper and evaluating their results for myself.

Abstract: Pathfinding in uniform-cost grid environments is a problem commonly found in application areas such as robotics and video games. The state-of-the-art is dominated by hierarchical pathfinding algorithms which are fast and have small memory overheads but usually return suboptimal paths. In this paper we present a novel search strategy, specific to grids, which is fast, optimal and requires no memory overhead. Our algorithm can be described as a macro operator which identifies and selectively expands only certain nodes in a grid map which we call jump points. Intermediate nodes on a path connecting two jump points are never expanded. We prove that this approach always computes optimal solutions and then undertake a thorough empirical analysis, comparing our method with related works from the literature. We find that searching with jump points can speed up A* by an order of magnitude and more and report significant improvement over the current state of the art.

URL: Download the paper at this location
Alternate site: pdf

PRESENTATION:
  pdf file

Paper: "A self-localization and path planning technique for mobile robot navigation,"
Jia-Heng Zhou and Huei-Yung Lin.
In Proceedings of the 9th World Congress on Intelligent Control and Automation (WCICA 2009), 2009.

Reason for choice: I found a marvelous paper on mobile robot path planning that was quite insightful especially when it described the heuristics they used to prune the state space (quite significantly I might add). I believe the paper is relevant and perhaps even interesting to many of the students in the class who used robots for their projects. Generally speaking, the paper demonstrates a fusion of a number of algorithms which should provide me with plenty of material to report on in depth.

Abstract: In this paper, we propose a system to cope with the problem of autonomous mobile robot navigation. It is able to perform path planning and localize the robot in the real world environment. The path planning is first carried out using the known map, and the laser range scanner is then used to localize the robot based on the ICP registration technique. During the robot motion, the potential field is taken into account for obstacle avoidance. For the path planning, the visibility graph is established based on the current position of the robot. The Dijkstra algorithm is then used to find the shortest path to the goal position. Experimental results for both the simulation and real world environment are presented.

URL: Download the paper at this location

PRESENTATION:
  pdf file

Paper: "System modelling and online optimal management of MicroGrid using Mesh Adaptive Direct Search,"
Faisal A. Mohamed and Heikki N. Koivo.
Journal of Electrical Power and Energy Systems, June 2010, 32(5):398--407.

Reason for choice: This paper was chosen because my research area is power optimization, and find this paper interesting in learning more advanced techniques that can be used in future research.

Abstract: This paper presents a generalized formulation to determine the optimal operating strategy and cost optimization scheme for a MicroGrid. Prior to the optimization of the MicroGrid itself, models for the system components are determined using real data. The proposed cost function takes into consideration the costs of the emissions, NO_x, SO₂, and CO₂, start-up costs, as well as the operation and maintenance costs. A daily income and outgo from sold or purchased power is also added. The MicroGrid considered in this paper consists of a wind turbine, a micro turbine, a diesel generator, a photovoltaic array, a fuel cell, and a battery storage. In this work, the Mesh Adaptive Direct Search (MADS) algorithm is used to minimize the cost function of the system while constraining it to meet the customer demand and safety of the system. In comparison with previously proposed techniques, a significant reduction is obtained.

URL: Download the paper at this location

PRESENTATION:
  pdf file

Paper: "Optiml metric planning with state sets in automata representation,"
Bjorn Ulrich Borowsky and Stefan Edelkamp.
In Proceedings of the Twenty-Third AAAI Conference on Artificial Intelligence (AAAI-2011), pages 874--879, 2008.

Reason for choice: Presburger arithmetic is a fully decidable theory of integers over addition. This paper involves planning, but I know that some model checkers use Presburger formulas to model operations on machine integers (neglecting overflow, but a good cost tradeoff). My research does involve variables of parametrized domain size (like Dijkstra's token ring: for convergence, N processes need variables of size N+1 at least), so knowing how infinite-state systems are handled should be useful.

Abstract: This paper proposes an optimal approach to infinite-state action planning exploiting automata theory. State sets and actions are characterized by Presburger formulas and represented using minimized finite state machines. The exploration that contributes to the planning via model checking paradigm applies symbolic images in order to compute the deterministic finite automaton for the sets of successors. A large fraction of metric planning problems can be translated into Presburger arithmetic, while derived predicates are simply compiled away. We further propose three algorithms for computing optimal plans; one for uniform action costs, one for the additive cost model, and one for linear plan metrics. Furthermore, an extension for infinite state sets is discussed.

URL: Download the paper at this location

PRESENTATION:
  pdf file

Paper: "Dynamic Game Difficulty Scaling Using Adaptive Behavior-Based AI,"
Chin Hiong Tan, Kay Chen Tan, and Arthur Tay. IEEE Transactions on Computational Intelligence and AI in Games, 2011, 3(4):289--301.

Reason for choice: I chose this paper because I am interested in AI as it applies to video games. This paper is not focused on developing an AI that will necessarily win the game in an optimal way. Instead, it focuses on developing an AI for nonplayer characters that is satisfying for a human to play against by providing them with a sufficient challenge for their skill level. This is something many players desire in a modern video game

Abstract: Games are played by a wide variety of audiences. Different individuals will play with different gaming styles and employ different strategic approaches. This often involves interacting with nonplayer characters that are controlled by the game AI. From a developer's standpoint, it is important to design a game AI that is able to satisfy the variety of players that will interact with the game. Thus, an adaptive game AI that can scale the difficulty of the game according to the proficiency of the player has greater potential to customize a personalized and entertaining game experience compared to a static game AI. In particular, dynamic game difficulty scaling refers to the use of an adaptive game AI that performs game adaptations in real time during the game session. This paper presents two adaptive algorithms that use ideas from reinforcement learning and evolutionary computation to improve player satisfaction by scaling the difficulty of the game AI while the game is being played. The effects of varying the learning and mutation rates are examined and a general rule of thumb for the parameters is proposed. The proposed algorithms are demonstrated to be capable of matching its opponents in terms of mean scores and winning percentages. Both algorithms are able to generalize well to a variety of opponents.

URL: Download the paper at this location

PRESENTATION:
  pdf file

Paper: "Autonomous Driving in Urban Environments: Boss and the Urban Challenge,"
Chris Urmson et al.
Journal of Field Robotics Special Issue: Special Issue on the 2007 DARPA Urban Challenge Intelligence Part I, August 2008, 25(8):425-466,

Reason for choice: This summer, I become very interested in self-driving and auto piloting when I saw Google self-driving car's test running in California. As of September 2012, three U.S. states (Nevada, Florida and California) have passed laws permitting Google's self-driving cars, that means the self driving car will become a part of our daily life in the near future. This paper is newly published and talks about the key technologies in this application. I think the research in self-driving car technology will advance our knowledge in real applications of AI.

Abstract: Boss is an autonomous vehicle that uses on-board sensors (global positioning system, lasers, radars, and cameras) to track other vehicles, detect static obstacles, and localize itself relative to a road model. A three-layer planning system combines mission, behavioral, and motion planning to drive in urban environments. The mission planning layer considers which street to take to achieve a mission goal. The behavioral layer determines when to change lanes and precedence at intersections and performs error recovery maneuvers. The motion planning layer selects actions to avoid obstacles while making progress toward local goals. The system was developed from the ground up to address the requirements of the DARPA Urban Challenge using a spiral system development process with a heavy emphasis on regular, regressive system testing. During the National Qualification Event and the 85-km Urban Challenge Final Event, Boss demonstrated some of its capabilities, qualifying first and winning the challenge

URL: Download the paper at this location

PRESENTATION:
  pptx file
  pdf file

Paper: "Distributed Quality-of-Service routing of best constrained shortest paths, "
Abdelhamid Mellouk, Said Hoceini, Farid Baguenine, Mustapha Cheurfa.
In Proceedings of the IEEE Symposium on Computers and Communications (ISCC-2008), pages 915--919, 2008.

Reason for choice: I am interested in communication and computer networks. This paper explains the algorithm for finding the shortest path for routing algorithms. The routing algorithm and finding the best path in a communication network are the keys to the success of future high communication networks.

Abstract: High speed modern communication networks are required to integrate and support multimedia application which requires differentiated quality-of-service (QoS) guarantees. Routing mechanism is a key to success of future communication networking. However, it is often complicated by the notion of guaranteed QoS, which can either be related to time, cost, packet loss or bandwidth requirements. Communication network requires that as the load levels, traffic patterns and topology of the network change, the routing policy also adapts. In this paper, we present a QoS based routing to construct dynamic state-dependent routing policies. The proposed algorithm used a reinforcement learning paradigm to optimize two QoS criteria: cumulative cost path based on hop count and end-to-end delay. Multiple paths are searched in parallel to find the N best qualified ones. In order to improve the overall network performance, a load balancing policy is defined and depends on a dynamical traffic path probability distribution function. The performance of our algorithm for different levels of traffic load is compared experimentally with standard optimal path routing algorithms for the same problem.

URL: Download the paper at this location

PRESENTATION:
  pptx file
  pdf file

Paper: "3D Environment Reconstruction Using Modified Color ICP Algorithm by Fusion of a Camera and 3D Laser Range Finder,"
Ji Hoon Joung.
In Proceedings of the 2009 IEEE/RSJ International Conference on Intelligent Robots and Systems , 2009.

Reason for choice: I am researching the indoor UAV Navigation System. This is a difficult the problem because GPS cannot work in the room. I have to use other absolute positioning such as vision to correct the error of the inertial navigation system. The paper introduces a method to get an accurate and stable estimation 3D environment data robustly while the robot is moving. We can use the data to locate the UAV.

Abstract: In this paper, we propose a system which reconstructs the environment with both color and 3D information. We perform extrinsic calibration of a camera and a LRF (Laser Range Finder) to fuse 3D information and color information of objects. We also formularize an equation to measure the result of the calibration. Moreover, we acquire 3D data by rotating 2D LRF with camera, and use ICP (Iterative Closest Point) algorithm to combine data acquired in other places. We use the SIFT (Scale Invariant Feature Transform) matching for the initial estimation of ICP algorithm. It offers accurate and stable initial estimation robust to motion change compare to odometer. We also modify the ICP algorithm using color information. Computation time of ICP algorithm can be reduced by using color information.

URL: Download the paper at this location

PRESENTATION:
  pptx file
  pdf file

Paper: "Course of Action Generation for Cyber Security Using Classical Planning,"
Mark Boddy, Johnathan Gohde, Tom Haigh, Steven Harp.

Reason for choice: I chose this paper because I am doing a research on Cyber security of Power Substations. I am looking at different techniques to analyze cyber vulnerabilities in a substation network. This paper demonstrates that classical planning can be a good approach for this problem. Although this paper does not describe the complete solution to the problem of cyber security, it lays foundation for further research.

Abstract: We report on the results of applying classical planning techniques to the problem of analyzing computer network vulnerabilities. Specifically, we are concerned with the generation of Adversary Courses of Action, which are extended sequences of exploits leading from some initial state to an attacker’s goal. In this application, we have demonstrated the generation of attack plans for a simple but realistic web-based document control system, with excellent performance compared to the prevailing state of the art in this area.
In addition to the new capabilities gained in the area of vulnerability analysis, this implementation provided some insights into performance and modeling issues for classical planning systems, both specifically with regard to METRIC-FF and other forward heuristic planners, and more generally for classical planning. To facilitate additional work in this area, the domain model on which this work was done will be made freely available. See the paper’s Conclusion for details.

URL: Download the paper at this location

PRESENTATION:
  pptx file
  pdf file

Paper: "Improved Human-Robot Team Performance Using Chaski, A Human-Inspired Plan Execution System,"
Julie Shah, James Wiken, Brian Williams, and Cynthia Breazeal.
In Proceedings of the 6th ACM/IEEE International Conference on Human-Robot Interaction (HRI 2011), pages 29-36, 2011.

Reason for choice: I have been always fascinated by robots and their communication with the real world. The idea of learning of robots from humans is interesting. This will help the robot to customize and adapt itself according to current environment and user.

Abstract: We describe the design and evaluation of Chaski, a robot plan execution system that uses insights from human-human teaming to make human-robot teaming more natural and fluid. Chaski is a task-level executive that enables a robot to collaboratively execute a shared plan with a person. The system chooses and schedules the robot’s actions, adapts to the human partner, and acts to minimize the human’s idle time.
We evaluate Chaski in human subject experiments in which a person works with a mobile and dexterous robot to collaboratively assemble structures using building blocks. We measure team performance outcomes for robots controlled by Chaski compared to robots that are verbally commanded, step-by-step by the human teammate. We show that Chaski reduces the human’s idle time by 85%, a statistically significant difference. This result supports the hypothesis that human-robot team performance is improved when a robot emulates the effective coordination behaviors observed in human teams

URL: Download the paper at this location

PRESENTATION:
  pptx file
  pdf file

Paper: "Can Cloud Computing be Used for Planning? An Initial Study,"
Qiang Lu, You Xu, Ruoyun Huang, Yixin Chen, and Guoliang Chen.
In Proceedings of the 3rd IEEE International Conference on Cloud Computing Technology and Science (CloudCom-11), 2011.

Reason for choice: Cloud computing has become very popular recently and is one of my research topics. In addition, planning is one of the main subfields of artificial intelligence. Both cloud computing and search techniques for planning are discussed in this paper. It will be interesting to study whether could computing can be used in planning and result in some advantages.

Abstract: Cloud computing is emerging as a prominent computing model. It provides a low-cost, highly accessible alternative to other traditional high-performance computing platforms. It also has many other benets such as high availability, scalability, elasticity, and free of maintenance. Given these attractive features, it is very desirable if automated planning can exploit the large, affordable computational power of cloud computing. However, the latency in inter-process communication in cloud computing makes most exist- ing parallel planning algorithms unsuitable for cloud computing. In this paper, we propose a portfolio stochastic search framework that takes advantage of and is suitable for cloud computing. We first study the running time distribution of MonteCarlo Random Walk (MRW) search, a stochastic planning algorithm, and show that the running time distribution usually has remarkable variability. Then, we propose a portfolio search algorithm that is suitable for cloud computing, which typically has abundant computing cores but high communication latency between cores. Further, we introduce an enhanced portfolio with multiple parameter settings to improve the efciency of the algorithm. We implement the portfolio search algorithm in both a local cloud and the Windows Azure cloud. Experimental results show that our algorithm achieves good, in many cases superlinear, speedup in the cloud platforms. Moreover, our algorithm greatly reduces the running time variance of the stochastic search and improves the solution quality. We also show that our scheme is economically sensible and robust under processor failures.

URL: Download the paper at this location

PRESENTATION:
  pptx file
  pdf file

Paper: "Recognizing Human Activities User-independently on Smartphones Based on Accelerometer Data,"
Pekka Siirtola and Juha Roning.
International Journal of Interactive Multimedia and Artificial Intelligence, June, 1(5):38-45.

Reason for choice: I am interested in implementations getting artificial intelligence to the every day life. Smartphones have became part of our every day life. I would like to see more context aware applications on smartphones.

Abstract: Real-time human activity recognition on a mobile phone is presented in this article. Unlike in most other studies, not only the data were collected using the accelerometers of a smartphone, but also models were implemented to the phone and the whole classification process (preprocessing, feature extraction and classification) was done on the device. The system is trained using phone orientation independent features to recognize five everyday activities: walking, running, cycling, driving a car and sitting/standing while the phone is in the pocket of the subject's trousers. Two classifiers were compared, knn (k nearest neighbors) and QDA (quadratic discriminant analysis). The models for real-time activity recognition were trained offline using a data set collected from eight subjects and these offline results were compared to real-time recognition rates, which are obtained by implementing models to mobile activity recognition application which currently supports two operating systems: Symbian^3 and Android. The results show that the presented method is light and, therefore, suitable for be used in real-time recognition. In addition, the recognition rates on the smartphones were encouraging, in fact, the recognition accuracies obtained are approximately as high as offline recognition rates. Also, the results show that the method presented is not an operating system dependent.

URL: Download the paper at this location

PRESENTATION:
  pdf file

Paper: "Multi-Robot Perimeter Patrol in Adversarial Settings,"
Noa Agmon, Sarit Kraus and Gal A. Kaminka.
In Proceedings of the 2008 IEEE International Conference on Robotics and Automation (ICRA-2008), pages 2339--2345, 2008.

Reason for choice: I chose this paper because I am interested in Robotics and hoping that it will give me more idea about AI in Robotics. It has also close to and in direct relation with the lectures we cover in class.

Abstract: This paper considers the problem of multi-robot patrol around a closed area with the existence of an adversary attempting to penetrate into the area. In case the adversary knows the patrol scheme of the robots and the robots use a deterministic patrol algorithm, then in many cases it is possible to penetrate with probability 1. Therefore this paper considers a non-deterministic patrol scheme for the robots, such that their movement is characterized by a probability p. This patrol scheme allows reducing the probability of penetration, even under an assumption of a strong opponent that knows the patrol scheme. We offer an optimal polynomial-time algorithm for finding the probability p such that the minimal probability of penetration detection throughout the perimeter is maximized. We describe three robotic motion models, defined by the movement characteristics of the robots. The algorithm described herein is suitable for all three models.

URL: Download the paper at this location

PRESENTATION:
  pptx file
  pdf file

Paper: "Thresholded Rewards: Acting Optimally in Timed, Zero-Sum Games,"
Colin McMillen and Manuela Veloso.
In Proceedings of the Twenty-Second AAAI Conference on Artificial Intelligence (AAAI-2007), pages 1250--1255, 2007.

Reason for choice: I am very curious about AI application in games. And this one is about maximizing the winning chance, which is more exciting. While I was searching for papers, I noticed that a good amount of papers mentioned Markov Decision Processes. I am not sure whether it is very closely related to the decision-making or planning strategies we were introduced to in the class, but this one became interesting to me since it is used very often. So I believe this is a part that I should learn more about.

Abstract: In timed, zero-sum games, the goal is to maximize the probability of winning, which is not necessarily the same as maximizing our expected reward. We consider cumulative intermediate reward to be the difference between our score and our opponent's score; the "true" reward of a win, loss, or tie is determined at the end of a game by applying a threshold function to the cumulative intermediate reward. We introduce thresholded-rewards problems to capture this dependency of the final reward outcome on the cumulative intermediate reward. Thresholded-rewards problems reflect different real-world stochastic planning domains, especially zero-sum games, in which time and score need to be considered. We investigate the application of thresholded rewards to finite-horizon Markov Decision Processes (MDPs). In general, the optimal policy for a thresholded-rewards MDP will be nonstationary, depending on the number of time steps remaining and the cumulative intermediate reward. We introduce an efficient value iteration algorithm that solves thresholded-rewards MDPs exactly, but with running time quadratic on the number of states in the MDP and the length of the time horizon. We investigate a number of heuristic-based techniques that efficiently find approximate solutions for MDPs with large state spaces or long time horizons.

URL: Download the paper at this location

PRESENTATION:
  pdf file

Paper: "Position Estimation by Registration to Planetary Terrain,"
Aashish Sheshadri, Kevin M. Peterson, Heather L. Jones and William L. "Red" Whittaker
In Proceedings of the International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI 2012) September 13-15, 2012. Hamburg, Germany , pages 432--438, 2012.

Reason for choice:

Abstract: LIDAR-only and camera-only approaches to global localization in planetary environments have relied heavily on availability of elevation data. The low-resolution nature of available DEMs limits the accuracy of these methods. Availability of new high-resolution planetary imagery motivates the rover localization method presented here. The method correlates terrain appearance with orthographic imagery. A rover generates a colorized 3D model of the local terrain using a panorama of camera and LIDAR data. This model is orthographically projected onto the ground plane to create a template image. The template is then correlated with available satellite imagery to determine rover location. No prior elevation data is necessary. Experiments in simulation demonstrate 2m accuracy. This method is robust to 30 degree differences in lighting angle between satellite and rover imagery.

URL: Download the paper at this location

PRESENTATION:
  pdf file

Paper: "ANA*: Anytime Nonparametric A*,"
Jur van den Berg, Rajat Shah, Arthur Huang, and Ken Goldberg.
In Proceedings of the Twenty-Fifth AAAI Conference on Artificial Intelligence (AAAI-2011), pages 105--111, 2011.

Reason for choice: One of my biggest interests in Artificial Intelligence is realtime pathfinding in robotics and games. An important solution for these problems is the use of anytime search algorithms to allow the agent to either accept a suboptimal solution if it requires an answer quickly, or to wait for a better solution if a decision is not required immediately. The advantages ANA* offers over other anytime algorithms are better automation (by eliminating the need for parameters), and more overall speed and optimality (shown through experimental results). This can become very important in situations where poor decision speed could cause damaging collisions, and where time is too limited to allow for agents to wait around for a better solution.

Abstract: Anytime variants of Dijkstra's and A* shortest path algorithms quickly produce a suboptimal solution and then improve it over time. For example, ARA* introduces a weighting value (ε) to rapidly find an initial suboptimal path and then reduces εepsilon; to improve path quality over time. In ARA*, εepsilon; is based on a linear trajectory with ad-hoc parameters chosen by each user. We propose a new Anytime A* algorithm, Anytime Nonparametric A* (ANA*), that does not require ad-hoc parameters, and adaptively reduces εepsilon; to expand the most promising node per iteration, adapting the greediness of the search as path quality improves. We prove that each node expanded by ANA* provides an upper bound on the suboptimality of the current-best solution. We evaluate the performance of ANA* with experiments in the domains of robot motion planning, gridworld planning, and multiple sequence alignment. The results suggest that ANA* is as efficient as ARA* and in most cases: (1) ANA* finds an initial solution faster, (2) ANA* spends less time between solution improvements, (3) ANA* decreases the suboptimality bound of the current-best solution more gradually, and (4) ANA* finds the optimal solution faster. ANA* is freely available from Maxim Likhachev’s Search-based Planning Library (SBPL).

URL: Download the paper at this location

PRESENTATION:
  pdf file