No description available.

No description available.

As an attempt to steer away from developing an autonomous robot with complex centralised intelligence, this thesis proposes an intelligent environment infrastructure where intelligences are distributed in the environment through collaborative vision sensors mounted in a physical architecture, forming a wireless sensor network, to enable the navigation of unintelligent robots within that physical architecture. The aim is to avoid the bottleneck of centralised robot intelligence that hinders the application and exploitation of autonomous robot. A bio-mimetic snake algorithm is proposed to coordinate the distributed vision sensors for the generation of a collision free Reference-snake (R-snake) path during the path planning process. By following the R-snake path, a novel Accompanied snake (A-snake) method that complies with the robot's nonholonomic constraints for trajectory generation and motion control is introduced to generate real time robot motion commands to navigate the robot from its current position to the target position. A rolling window optimisation mechanism subject to control input saturation constraints is carried out for time-optimal control along the A-snake. A comprehensive simulation software and a practical distributed intelligent environment with vision sensors mounted on a building ceiling are developed. All the algorithms proposed in this thesis are first verified by the simulation and then implemented in the practical intelligent environment. A model car with less on-board intelligence is successfully controlled by the distributed vision sensors and demonstrated superior mobility.

As an attempt to steer away from developing an autonomous robot with complex centralised intelligence, this thesis proposes an intelligent environment infrastructure where intelligences are distributed in the environment through collaborative vision sensors mounted in a physical architecture, forming a wireless sensor network, to enable the navigation of unintelligent robots within that physical architecture. The aim is to avoid the bottleneck of centralised robot intelligence that hinders the application and exploitation of autonomous robot. A bio-mimetic snake algorithm is proposed to coordinate the distributed vision sensors for the generation of a collision free Reference-snake (R-snake) path during the path planning process. By following the R-snake path, a novel Accompanied snake (A-snake) method that complies with the robot's nonholonomic constraints for trajectory generation and motion control is introduced to generate real time robot motion commands to navigate the robot from its current position to the target position. A rolling window optimisation mechanism subject to control input saturation constraints is carried out for time-optimal control along the A-snake. A comprehensive simulation software and a practical distributed intelligent environment with vision sensors mounted on a building ceiling are developed. All the algorithms proposed in this thesis are first verified by the simulation and then implemented in the practical intelligent environment. A model car with less on-board intelligence is successfully controlled by the distributed vision sensors and demonstrated superior mobility.

A training unit has been developed that allows NVESD researchers to develop training simulations within virtual environments to enhance infantry skill and awareness. A ground station was developed to house a computer, power system, and video transmission system. This station will allow for a remote operator to wirelessly send a video/audio stream to the handset. The ground station also allows the use of external video and audio inputs to be sent using onboard converters. Different wireless frequencies were evaluated to determine the best for long-range transmission of content. A handset was developed from a carbon fiber prototype shell. The handset features a video receiver, display, power system, OSD system, and external video inputs. The user can view transmitted video and audio while obtaining real-time GPS feedback from the OSD. The alternate video input allows the handset to be used within the virtual environments developed at the University of Kentucky’s Center for Visualization for virtual environments. This thesis will present the research conducted in order to develop Mark VIIE training unit including the requirements for the project, the desired functionality, the NVESD provided equipment, the analysis of the prospective components, the design of custom fabricated parts, and the assembly and integration of the components into a complete system.

Recently, the efforts involved in the digitization and digitalization of logistics processes have grown tremendously. In line with such efforts, we investigate the potential of the process-integrated measuring and load planning of large transport assets. More precisely, considering the case of a German timber processor and retailer, we implement a digital measuring system, which performs precise measuring of regularly shaped wooden assets. The cognitive system uses laser and vision sensors, and measurements can be performed during the asset’s transportation on a forklift. The resulting data can be used to conduct a comprehensive load planning for scheduled delivery tours. The performance of our measurement system is evaluated using a small example dataset of the use case at hand. The a-priori set goal of maximum deviations of 5 cm, 7 cm and 14 cm in height, width and length, respectively, are achieved in 89% of the test cases. The proposed load planning algorithm is integrated in a commercial tour planning service to verify the feasibility of serving several customers within the same tour. We present the method’s applicability to our described use case of integrated measurement and planning.

No description available.

Defects such as cracks, overlaps and impressions occur during the production of press-hardened car body components. At present, these types of defects are counteracted in the industrial environment by costly visual inspections carried out by humans. Due to the poor efficiency of visual inspection compared to automated inspection and the risk of defects not being detected, the use of AI-based smart vision sensors is being evaluated in order to enable an automated component inspection process with their help. For the realisation of the test, the most relevant defect types deformation, crack and overlap are identified using a Pareto analysis.

Bei der Herstellung pressgehärteter Karosseriebauteile treten Fehler wie Risse, Überlappungen und Abdrücke auf. Gegenwärtig wird diesen Fehlerarten im industriellen Umfeld durch kostenaufwändige, von Menschen durchgeführte Sichtkontrollen entgegengewirkt. Aufgrund des schlechten Wirkungsgrades der visuellen Prüfung gegenüber einer automatisierten Prüfung und der Gefahr des Nichterkennens von Fehlern, wird der Einsatz von KI-basierten Smart-Vision-Sensoren evaluiert, um mit deren Hilfe einen automatisierten Bauteilprüfprozess zu ermöglichen.

[ES] Los robots paralelos (PR por sus siglas en inglés) son mecanismos donde el efector final está unido a la base, mediante al menos dos cadenas cinemáticas abiertas. Los PRs ofrecen una gran capacidad de carga y alta precisión, lo que los hace adecuados para diversas aplicaciones, entre ellas la interacción persona-robot. Sin embargo, en las proximidades de una singularidad Tipo II (singularidad dentro del espacio de trabajo), un PR pierde el control sobre los movimientos del efector final. La pérdida de control representa un riesgo importante para los usuarios, especialmente en rehabilitación robótica. En las últimas décadas, los PR se han popularizado en la rehabilitación de miembros inferiores debido al aumento del número de personas que viven con limitaciones físicas. Así, esta tesis trata sobre la detección y evitación de singularidades de Tipo II para asegurar total control de un PR no redundante para la rehabilitación y diagnóstico de rodilla, denominado 3UPS+RPU. En la literatura, existen varios índices para detectar y medir la cercanía a una singularidad basados en métodos analíticos y geométricos. Sin embargo, algunos de estos índices carecen de significado físico y son incapaces de identificar los actuadores responsables de la pérdida de control. Esta tesis aporta dos novedosos índices para detectar y medir la proximidad a una singularidad de Tipo II, capaces de identificar el par de actuadores responsables de la singularidad. Los dos índices son los ángulos entre los componentes lineal (T_i,j) y angular (O_i,j) de dos Twist Screw de Salida (OTS por sus siglas en inglés) normalizados i,j. Una singularidad Tipo II es detectada cuando T_i,j = O_i,j = 0 y su proximidad se mide mediante los mínimos ángulos T_i,j (minT) y O_i,j (minO) para los casos plano y espacial, respectivamente. La eficacia de los índices T_i,j y O_i,j se evalúa de forma teórica y experimental en un robot 3UPS+RPU y un mecanismo de cinco barras. Además, se propone un procedimiento experimental para el adecuado establecimiento del límite de cercanía a una singularidad de Tipo II mediante la aproximación progresiva del PR a una singularidad y la medición de la última posición controlable. Posteriormente, se desarrollan dos nuevos algoritmos deterministas para liberar y evitar una singularidad de Tipo II basados en minT y minO para PR no redundantes. minT y minO se utilizan para identificar los dos actuadores a mover para liberar o evitar el PR de una singularidad. Ambos algoritmos requieren una medición precisa de la pose alcanzada por el efector final. El algoritmo para liberar un PR de una configuración singular se aplica con éxito en un controlador híbrido basado en visión artificial para el PR 3UPS+RPU. El controlador utiliza un sistema de fotogrametría para medir la pose del robot debido a la degeneración del modelo cinemático en las proximidades de una singularidad. El algoritmo de evasión de singularidades Tipo II se aplica a la planificación offline y online de trayectorias no singulares para un mecanismo de cinco barras y el PR 3UPS+RPU. Estas aplicaciones verifican el bajo coste computacional y la mínima desviación introducida en la trayectoria original por los nuevos algoritmos. La implementación directa de un controlador de fuerza/posición en el PR 3UPS+RPU es insegura porque el paciente podría llevar involuntariamente al PR a una singularidad. Por lo tanto, esta tesis concluye presentando un novedoso controlador de fuerza/posición complementado con el algoritmo de evasión de singularidades de Tipo II. El nuevo controlador se evalúa durante rehabilitación activa de una pierna de maniquí y una pierna humana no lesionada. Los resultados muestran que el nuevo controlador combinado mantiene el PR 3UPS+RPU lejos de configuraciones singulares con una desviación mínima de la trayectoria original. Por lo tanto, esta tesis habilita el 3UPS+RPU PR para la rehabilitación segura de miembros inferiores lesionados. / [CAT] Els robots paral·lels (PR per les seues sigles en anglés) són mecanismes on l'efector final està unit a la base, mitjançant almenys dues cadenes cinemàtiques obertes. Els PRs ofereixen una gran capacitat de càrrega i alta precisió, la qual cosa els fa adequats per a diverses aplicacions, entre elles la interacció persona-robot. No obstant això, en les proximitats d'una singularitat Tipus II (singularitat dins de l'espai de treball), un PR perd el control sobre els moviments de l'efector final. La pèrdua de control representa un risc important per als usuaris, especialment en rehabilitació robòtica. En les últimes dècades, els PR s'han popularitzat en la rehabilitació de membres inferiors a causa de l'augment del nombre de persones que viuen amb limitacions físiques. Així, aquesta tesi tracta sobre la detecció i evació de singularitats de Tipus II per a assegurar total control d'un PR no redundant per a la rehabilitació i diagnòstic de genoll, denominat 3UPS+RPU. En la literatura, existeixen diversos índexs per a detectar i mesurar la proximitat a una singularitat basats en mètodes analítics i geomètrics. No obstant això, alguns d'aquests índexs manquen de significat físic i són incapaços d'identificar els actuadors responsables de la pèrdua de control. Aquesta tesi aporta dos nous índexs per a detectar i mesurar la proximitat a una singularitat de Tipus II, capaços d'identificar el parell d'actuadors responsables de la singularitat. Els dos índexs són els angles entre els components lineal (T_i,j) i angular (O_i,j) de dues Twist Screw d'Eixida (OTS per les seues sigles en engonals) normalitzats i,j. Una singularitat Tipus II és detectada quan T_i,j = O_i,j = 0 i la seua proximitat es mesura mitjançant els minimos angles T_i,j (minT) i O_i,j (minO) per als casos pla i espacial, respectivament. L'eficàcia dels índexs T_i,j i O_i,j es evalua de manera teòrica i experimental en un robot 3UPS+RPU i un mecanisme de cinc barres. A més, es proposa un procediment experimental per a l'adequat establiment del límit de proximitat a una singularitat de Tipus II mitjançant l'aproximació progressiva del PR a una singularitat i el mesurament de l'última posició controlable. Posteriorment, es desenvolupen dos nous algorismes deterministes per a alliberar i evadir una singularitat de Tipus II basats en minT i minO per a PR no redundants. minT i minO s'utilitzen per a identificar els dos actuadors a moure per a alliberar o evadir el PR d'una singularitat. Aquests algorismes requereixen un mesurament precís de la posa aconseguida per l'efector final. L'algorisme per a alliberar un PR d'una configuració singular s'aplica amb èxit en un controlador híbrid basat en visió artificial per al PR 3UPS+RPU. El controlador utilitza un sistema de fotogrametria per a mesurar la posa del robot a causa de la degeneració del model cinemàtic en les proximitats d'una singularitat. L'algorisme d'evació de singularitats Tipus II s'aplica a la planificació offline i en línia de trajectòries no singulars per a un mecanisme de cinc barres i el PR 3UPS+RPU. Aquestes aplicacions verifiquen el baix cost computacional i la mínima desviació introduïda en la trajectòria original pels nous algorismes. La implementació directa d'un controlador de força/posició en el PR 3UPS+RPU és insegura perquè el pacient podria portar involuntàriament al PR a una singularitat. Per tant, aquesta tesi conclou presentant un nou controlador de força/posició complementat amb l'algorisme d'evació de singularitats de Tipus II. El nou controlador s'avalua durant la rehabilitació activa d'una cama de maniquí i una cama humana no lesionada. Els resultats mostren que el nou controlador combinat manté el PR 3UPS+RPU lluny de configuracions singulars amb una desviació mínima de la trajectòria original. Per tant, aquesta tesi habilita el 3UPS+RPU PR per a la rehabilitació segura dels membres inferiors lesionats. / [EN] Parallel Robots (PR)s are mechanisms where the end-effector is linked to the base by at least two open kinematics chains. The PRs offer a high payload and high accuracy, making them suitable for various applications, including human robot interaction. However, in proximity to a Type II singularity (singularity within the workspace), a PR loses control over the movements of the end-effector. The loss of control represents a major risk for users, especially in robotic rehabilitation. In the last decades, PRs have become popular in lower limb rehabilitation because of the increment in the number of people living with physical limitations. Thus, this thesis is about the detection and avoidance of Type II singularities to ensure complete control of a non-redundant PR for knee rehabilitation and diagnosis named 3UPS+RPU. In the literature, several indices exist to detect and measure the closeness to a singular configuration based on analytical and geometrical methods. However, some of these indices have no physical meaning, and they are unable to identify the actuators responsible for the loss of control. This thesis contributes two novel indices to detect and measure the proximity to a Type II singularity capable of identifying the pair of actuators responsible for the singularity. The two indices are the angles between the linear (T_i,j) and the angular (O_i,j) components of two i,j normalised Output Twist Screws (OTSs). A Type II singularity is detected when the angles T_i,j = O_i,j = 0 and its closeness is measured by the minimum T_i,j (minT) and minimum O_i,j (minO) for planar and spatial cases, respectively. The effectiveness of the indices T_i,j and O_i,j is evaluated from a theoretical and experimental perspective in a 3UPS+RPU and a five bars mechanism. Moreover, an experimental procedure is proposed for setting a proper limit of closeness to a Type II singularity by the progressive approach of the PR to singular configuration and measuring the last controllable pose. Subsequently, two novel deterministic algorithms for releasing and avoiding Type II singularities based on minT and minO are developed for non-redundant PRs. The minT and minO are used to identify the two actuators to move for release or prevent the PR from the singularity. Both algorithms require an accurate measuring of the pose reached by the end-effector. The algorithm to release a PR from a singular configuration is successfully applied in a vision-based hybrid controller for the 3UPS+RPU PR. The controller uses a photogrammetry system to measure the pose of the robot due to the degeneration of the kinematic model in the vicinity of a singularity. The Type II singularity avoidance algorithm is applied to offline and online free-singularity trajectory planning for a five-bar mechanism and the 3UPS+RPU PR. These applications verify the low computation cost and the minimum deviation introduced in the original trajectory for both novel algorithms. The direct implementation of a force/position controller in the 3UPS+RPU PR is unsafe because the patient could unintentionally drive the PR to a Type II singularity. Therefore, this thesis concludes by presenting a novel force/position controller complemented with the Type II singularity avoidance algorithm. The complemented controller is evaluated during patient-active exercises in a mannequin leg and an uninjured human limb. The results show that the novel combined controller keeps the 3UPS+RPU PR far from singular configurations with a minimum deviation on the original trajectory. Hence, this thesis enables the 3UPS+RPU PR for the safe rehabilitation of injured lower limbs. / Pulloquinga Zapata, JL. (2023). A New Index for Detecting and Avoiding Type II Singularities for the Control of Non-Redundant Parallel Robots [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/194271