Abstract

In the field of active perception, object search is a widely studied problem. To search for an object in large rooms, it would be expensive to explore and check each object's similarity with the object of interest. The expense could uncontrollably bloat as the number of objects to be searched increases. If the objects are of the order of a 2-5cm, they appear very small, making it difficult for the present algorithms to recognize them. A general human strategy in such cases is to sparsely identify, from far away (4-6m), if the object of interest is present in the scene. Subsequently, each of the possible objects is analysed from closer proximity to recognize, for further manipulation. In this work, we present a similar framework. We reduce search-space, by identifying existential probability of a small object from a distance followed by a closer 3-D analysis of its point cloud to accurately recognize it. This is achieved by 2-D modelling of the objects using Gaussian Mixture Models followed by recognizing objects using efficient RGB-Depth based algorithm.

Abstract

The current paper proposes a trajectory optimization approach for navigating a non-holonomic wheeled mobile robot in dynamic environments. The dynamic obstacle's motion is not known and hence is represented by a band of predicted trajectories. The trajectory optimization can account for large number of predicted obstacle trajectories and seeks to avoid each predicted trajectory of every obstacle in the sensing range of the robot. The two primary contributions of the proposed trajectory optimization are (1): A computationally efficient method for computing the intersection space of collision avoidance constraints of large number of predicted obstacle trajectories. (2): A optimization framework to connect the current state to the solution space in time optimal fashion. The intersection/solution space computation is build on our earlier proposed concept of time scaled collision cone, which can be solved in closed form to obtain a set of formulae. These formulae describe how much and in what manner the temporal specification of a trajectory needs to be changed to avoid a given set of dynamic obstacles. This allows us to quickly evaluate solution space of time scaled collision cone over various candidate trajectories, thus reducing the problem of computing the intersection space to that of generating multiple homotopic trajectories. The optimization framework used to connect the current state to the solution space in time optimal fashion is based on the concept of non-linear time scaling, which induces a difference of convex form structure. Thus, on the theoretical side, we show that the various components of the proposed framework are computationally simple and involves solving sets of linear equations and using state of the art convex programming techniques. On the practical side we show that the proposed planner performs better than sampling based planners which treat dynamic obstacles as static over a short duration of time.

Abstract

In indoor environments, there would be several small objects lying around on the floor. In this work, we develop an efficient strategy to search for a set of queried objects amongst a large number of small objects lying around. Small objects of the order of 1cm – 5cm, appear very small, making it difficult for the present algorithms to recognize them from far away. A human like strategy in such cases is to infer each object's similarity to the queried objects, from far away. Subsequently, the objects of interest are approached and analyzed from a closer proximity through an optimal plan. We develop an optimal plan for the robot, to strategically visit a selected few among all the objects. From far away, we assign Existential Probabilities to the objects, indicating their similarity to queried objects. A Bayes' Net is constructed over the probabilities, to overlay and orient a Viewpoint Object Potential(VOP) map over potential search objects. VOP quantifies the probability of accurately recognizing an object through its RGB-D Point Cloud at various viewpoints. The belief from the Bayes' Net and the discriminative viewpoints from the VOP are utilized to formulate a Decision Tree which helps in building an optimal control plan. Hence, the robot reaches strategic viewpoints around potential objects, to recognize them through their RGB-D point clouds. The framework is experimentally evaluated using Kinect mounted on a Turtlebot using ROS platform.

Abstract

Frontier detection is a critical component in indoor mobile robot exploration, wherein the robot decides the next best location to move in order to continue with its mapping process. All frontier detection algorithms to the best of our knowledge require 3D locations of occupied regions as its input. In a monocular setting this entails a backend VSLAM algorithm that reconstructs the scene as the robot moves. Most monocular SLAM algorithms however provide sparse scene reconstruction from which frontiers cannot be reliably detected and estimated. In this effort we provide an alternate method of detecting frontiers during the course of robot motion that circumvents the requirement of dense mapping. Based on the observation that frontiers typically occur around vertical edges of walls, doors or tables we propose a novel linear chain CRF formulation that is able to detect the presence or absence of such frontier regions around such vertical edges. We used cues like increase in number of ground plane pixels and change in the spreading of optical flow vector, around those vertical edges. We also demonstrate that this method gives us more relevant frontiers as compared to methods based on reconstructing the scene through state-of-the art such SLAM algorithms such as PTAM. Finally, we present results in indoor scenes wherein frontiers are reliably detected around wall edges leading to new corridors, door edges leading to new rooms or corridors and table edges that opens up to a new space in rooms.

Abstract

Detecting multiple planes in images is a challenging problem, but one with many applications. Recent work such as J-Linkage and Ordered Residual Kernels have focussed on developing a domain independent approach to detect multiple structures. These multiple structure detection methods are then used for estimating multiple homographies given feature matches between two images. Features participating in the multiple homographies detected, provide us the multiple scene planes. We show that these methods provide locally optimal results and fail to merge detected planar patches to the true scene planes. These methods use only residues obtained on applying homography of one plane to another as cue for merging. In this paper, we develop additional cues such as local consistency of planes, local normals, texture etc. to perform better classification and merging. We formulate the classification as an MRF problem and use TRWS message passing algorithm to solve non metric energy terms and complex sparse graph structure. We show results on Michigan Indoor Corridor Dataset and our challenging dataset, common in robotics navigation scenarios. Experiments on the datasets demonstrate the accuracy of our plane detection relative to ground truth, with detailed comparisons to prior art.

Abstract

While the literature has been fairly dense in the areas of scene understanding and semantic labeling there have been few works that make use of motion cues to embellish semantic performance and vice versa. In this paper, we address the problem of semantic motion segmentation, and show how semantic and motion priors augments performance. We propose an algorithm that jointly infers the semantic class and motion labels of an object. Integrating semantic, geometric and optical flow based constraints into a dense CRF-model we infer both the object class as well as motion class, for each pixel. We found improvement in performance using a fully connected CRF as compared to a standard clique-based CRFs. For inference, we use a Mean Field approximation based algorithm. Our method outperforms recently proposed motion detection algorithms and also improves the semantic labeling compared to the state-of-the-art Automatic Labeling Environment algorithm on the challenging KITTI dataset especially for object classes such as pedestrians and cars that are critical to an outdoor robotic navigation scenario.

Abstract

For autonomous navigation in urban environments, the ability to detect road intersections in advance is crucial, especially in the absence of auxiliary geographic information. In this paper we investigate a 3D Point Cloud based solution for intersection recognition and road segment classification. We set up the intersection recognition problem as one of decoding a linear-chain Conditional Random Field (CRF). This allows us to encode temporal consistency relations between adjacent scans in our process, leading to a less error prone recognition algorithm. We quantify this claim experimentally. We first build a grid map of the point cloud, segmenting the region surrounding the robot into navigable and non-navigable regions. Then, based on our proposed beam model, we extract a descriptor of the scene. This we do as each scan is received from the robot. Based on the descriptor we build a linear chain-CRF. By decoding the CRF-chain we are able to recognize the type of road segment taken into consideration. With the proposed method, we are able to recognize Xjunctions, T-shaped intersections and standard non-branching road segments. We compare the CRF-based approach with a standard SVM based one and show performance gain due to the CRF formulation.

Abstract

The challenges in generating minimum time trajectory and control for generic quadrocopter flying through sophisticated and unknown environment are explored in this paper. The proposed method uses convex programming technique to optimize polynomial splines, which are numerically stable for high-order including large number of segments and easily formulated for efficient computation. Moreover, exploiting the differential flatness of system, these polynomial trajectories encode the dynamics and constraints of the vehicle and decouple them from trajectory planning. The framework is fast enough to be performed in real time and results in solution which is close to time optimal. As control inputs are computed from the generated trajectory in each update, they are applicable to achieve closed-loop control similar to model predictive controller.

Abstract

Today, IaaS cloud providers are dynamically minimizing the cost of data centers operations, while maintaining the Service Level Agreement (SLA). Currently, this is achieved by the live migration capability, which is an advanced state-of-the-art technology of Virtualization. However, existing migration techniques suffer from high network bandwidth utilization, large network data transfer, large migration time as well as the destination's VM failure during migration. In this paper, we propose Reliable Lazy Copy (RLC) - a fast, efficient and a reliable migration technique. RLC provides a reasonable solution for high-efficiency and less disruptive migration scheme by utilizing the three phases of the process migration. For effective network bandwidth utilization and reducing the total migration time, we introduce a learning phase to estimate the writable working set (WWS) prior to the migration, resulting in an almost single time transfer of the pages. Our approach decreases the total data transfer by 1.16 x - 12.21x and the total migration time by a factor of 1.42x - 9.84x against the existing approaches, thus providing a fast and an efficient, reliable VM migration of the VMs in the cloud.

Abstract

Multiprocessors have been widely used in modern high performance embedded system to meet the computational needs of smart, real time applications spread across multiple domains. While custom IPs (Intellectual Property) on FPGA based systems are commonly used, multiprocessing on FPGAs have not been explored enough due to concerns about meeting a right trade-off between area usage, achievable performance, and the required design time. This paper presents an implementation of a multiprocessing system on FPGA using multiple light weight soft processors (LWP) that work in conjunction with a custom hardware to achieve balanced performance to resource ratio. As an example we have implemented a TDEA (Triple Data Encryption Algorithm) cryptographic algorithm on our multiprocessing system that uses 2.5X less area, is 40% faster in performance, and requires 20% less power as compared to equivalent solution implemented using a commercial soft processor.