An Architecture for Human‐Guided Autonomy: Team TROOPER at the DARPA Robotics Challenge Finals

Recent robotics efforts have automated simple, repetitive tasks to increase execution speed and lessen an operator's cognitive load, allowing them to focus on higher‐level objectives. However, an autonomous system will eventually encounter something unexpected, and if this exceeds the tolerance of automated solutions, there must be a way to fall back to teleoperation. Our solution is a largely autonomous system with the ability to determine when it is necessary to ask a human operator for guidance. We call this approach human‐guided autonomy. Our design emphasizes human‐on‐the‐loop control where an operator expresses a desired high‐level goal for which the reasoning component assembles an appropriate chain of subtasks. We introduce our work in the context of the DARPA Robotics Challenge (DRC) Finals. We describe the software architecture Team TROOPER developed and used to control an Atlas humanoid robot. We employ perception, planning, and control automation for execution of subtasks. If subtasks fail, or if changing environmental conditions invalidate the planned subtasks, the system automatically generates a new task chain. The operator is able to intervene at any stage of execution, to provide input and adjustment to any control layer, enabling operator involvement to increase as confidence in automation decreases. We present our performance at the DRC Finals and a discussion about lessons learned.     

Visually Perceived Distance Judgments: Tablet-Based Augmented Reality Versus the Real World

Does visually perceived distance differ when objects are viewed in augmented reality (AR), as opposed to the real world? What are the differences? These questions are theoretically interesting, and the answers are important for the development of many tablet- and phone-based AR applications, including mobile AR navigation systems. This article presents a thorough literature review of distance judgment experimental protocols, and results from several areas of perceptual psychology. In addition to distance judgments of real and virtual objects, this section also discusses previous work in measuring the geometry of virtual picture space and considers how this work might be relevant to tablet AR. Then, the article presents the results of two experiments. In each experiment, observers bisected egocentric distances of 15 and 30 m in tablet-based AR and in the real world, in both indoor corridor and outdoor field environments. In AR, observers bisected the distances to virtual humans, while in the real world, they bisected the distances to real humans. This is the first reported research that directly compares distance judgments of real and virtual objects in a tablet AR system. Four key findings were: (1) In AR, observers expanded midpoint intervals at 15 m, but compressed midpoints at 30 m. (2) Observers were accurate in the real world. (3) The environmental setting—corridor or open field—had no effect. (4) The picture perception literature is important in understanding how distances are likely judged in tablet-based AR. Taken together, these findings suggest the depth distortions that AR application developers should expect with mobile and especially tablet-based AR.     

Template-based quadrilateral meshing

Generating quadrilateral meshes is a highly non-trivial task, as design decisions are frequently driven by specific application demands. Automatic techniques can optimize objective quality metrics, such as mesh regularity, orthogonality, alignment and adaptivity; however, they cannot make subjective design decisions. There are a few quad meshing approaches that offer some mechanisms to include the user in the mesh generation process; however, these techniques either require a large amount of user interaction or do not provide necessary or easy to use inputs. Here, we propose a template-based approach for generating quad-only meshes from triangle surfaces. Our approach offers a flexible mechanism to allow external input, through the definition of alignment features that are respected during the mesh generation process. While allowing user inputs to support subjective design decisions, our approach also takes into account objective quality metrics to produce semi-regular, quad-only meshes that align well to desired surface features.     

Template-based quadrilateral mesh generation from imaging data

This paper describes a novel template-based meshing approach for generating good quality quadrilateral meshes from 2D digital images. This approach builds upon an existing image-based mesh generation technique called Imeshp, which enables us to create a segmented triangle mesh from an image without the need for an image segmentation step. Our approach generates a quadrilateral mesh using an indirect scheme, which converts the segmented triangle mesh created by the initial steps of the Imesh technique into a quadrilateral one. The triangle-to-quadrilateral conversion makes use of template meshes of triangles. To ensure good element quality, the conversion step is followed by a smoothing step, which is based on a new optimization-based procedure. We show several examples of meshes generated by our approach, and present a thorough experimental evaluation of the quality of the meshes given as examples.     

Query-based learning for aerospace applications

Models of real-world applications often include a large number of parameters with a wide dynamic range, which contributes to the difficulties of neural network training. Creating the training data set for such applications becomes costly, if not impossible. In order to overcome the challenge, one can employ an active learning technique known as query-based learning (QBL) to add performance-critical data to the training set during the learning phase, thereby efficiently improving the overall learning/generalization. The performance-critical data can be obtained using an inverse mapping called network inversion (discrete network inversion and continuous network inversion) followed by oracle query. This paper investigates the use of both inversion techniques for QBL learning, and introduces an original heuristic to select the inversion target values for continuous network inversion method. Efficiency and generalization was further enhanced by employing node decoupled extended Kalman filter (NDEKF) training and a causality index (CI) as a means to reduce the input search dimensionality. The benefits of the overall QBL approach are experimentally demonstrated in two aerospace applications: a classification problem with large input space and a control distribution problem.     

Product-specific moisture levels: A conceptual framework

An improved schedule to evaluate surface-mount technology popcorn jeopardy based on physics of moisture absorption is proposed. Key features are: 1) each lettered level implies a certain performance irrespective of package thickness (i.e., no penalty for thinner packages); 2) each successive level has stepped moisture concentration at the interior (i.e., clear stepped discrimination between levels); and 3) the time required for preconditioning is substantially reduced from current practice and an acceleration factor can be clearly computed.     

Investigation and minimization of underfill delamination in flip chip packages

A generalized plane strain condition is assumed for an edge interfacial crack between die passivation and underfill on an organic substrate flip chip package. C4 solder bumps are explicitly modeled. Temperature excursions are treated as loading conditions. The design factors studied include underfill elastic modulus, underfill coefficient of thermal expansion (CTE), fillet height, and die overhang. Varying underfill modulus and CTE produces a different stress field at underfill/die passivation interface, different stress intensity factor (SIF), and phase angle (/spl psi/) even under the same loading condition. The baseline case uses underfill with elastic modulus of 6 GPa, CTE of 36 ppm//spl deg/C and fillet height equal to half die thickness. Four more cases involving underfill material properties are investigated by varying elastic modulus between 3 and 9 GPa, and by varying CTE between 26 and 46 ppm//spl deg/C. The effect of fillet height is also studied by assuming no fillet and full fillet, i.e., fillet height equal to die thickness. Finally, two cases concerning the influence of die overhang, defined as the nominal distance between outermost solder joint and die edge, are investigated. Fracture parameters, including energy release rate (G) and phase angle (/spl psi/), are evaluated as a function of dimensions. Underfill material properties (elastic modulus and CTE), fillet configuration, and die overhang can be optimized to reduce the risk of underfill delamination in flip chip or direct chip attach (DCA) applications.     

Fuse regrowth kinetics

The kinetics of fuse regrowth during temperature humidity bias (THB) stresses were modeled with a power law for relative humidity (exponent /spl sim/2.3) and a power law for bias voltage (exponent /spl sim/1.7). Using these results we determined the optimal laser cut parameters to maximize product reliability.     

Exact static element stiffness matrices of non‐symmetric thin‐walled curved beams

An evaluation procedure of exact static stiffness matrices for curved beams with non‐symmetric thin‐walled cross section are rigorously presented for the static analysis. Higher‐order differential equations for a uniform curved beam element are first transformed into a set of the first‐order simultaneous ordinary differential equations by introducing 14 displacement parameters where displacement modes corresponding to zero eigenvalues are suitably taken into account. This numerical technique is then accomplished via a generalized linear eigenvalue problem with non‐symmetric matrices. Next, the displacement functions of displacement parameters are exactly calculated by determining general solutions of simultaneous non‐homogeneous differential equations. Finally an exact stiffness matrix is evaluated using force–deformation relationships. In order to demonstrate the validity and effectiveness of this method, displacements and normal stresses of cantilever thin‐walled curved beams subjected to tip loads are evaluated and compared with those by thin‐walled curved beam elements as well as shell elements. Copyright © 2004 John Wiley & Sons, Ltd.     

High-Modulus Columns for Liquefaction Mitigation

This paper presents the performance of a shopping complex in Turkey where the soils were improved with jet-grout columns and preload fills and subjected to the 1999 Kocaeli Earthquake (M = 7.4). Under construction at the time of the earthquake, the Carrefour Shopping Center covers an area of 55,000?m2 and is founded on shallow footings, mats, and slabs-on-grade that rest on soft, saturated alluvial sediments consisting of clays, silts, and sands. High-modulus columns constructed by jet grouting were installed at close-to-moderate spacings to reduce anticipated static settlements in the clays and mitigate liquefaction in the sands. The site was subjected to a peak acceleration of approximately 0.2g during the earthquake. Grouting had been completed for about two-thirds of the site when the earthquake struck. Following the event, a field reconnaissance found stark contrast between the performance of the improved and unimproved sections. The jet-grout-treated areas suffered no apparent damage, whereas the unimproved sections of the complex, along with nearby untreated building sites, commonly suffered liquefaction-related settlements of up to 10 cm. This is the only case history known to the authors that documents the field performance of high-modulus columns used in this manner for liquefaction mitigation and direct instrumented measurement of liquefaction-induced settlements.