No description available.

Localization and mapping are vital capabilities for a mobile robot. These two capabilities strongly depend on each other and simultaneously executing both of these operations is called SLAM (Simultaneous Localization and Mapping). SLAM problem requires the environment to be represented with an abstract mapping model. It is possible to construct a map from point cloud of environment via scanner sensor systems. On the other hand, extracting higher level of features from point clouds and using these extracted features as an input for mapping system is also a possible solution for SLAM. In this work, a 4D feature based EKF SLAM system is constructed and open form of equations of algorithm are presented. The algorithm is able to use center of mass and direction of features as input parameters and executes EKF SLAM via these parameters. Performance of 4D feature based EKF SLAM was examined and compared with 3D EKF SLAM via monte-carlo simulations. By this way / it is believed that, contribution of adding a direction vector to 3D features is investigated and illustrated via graphs of monte-carlo simulations. At the second part of the work, a scanner sensor system with IR distance finder is designed and constructed. An algorithm was presented to extract planar features from data collected by sensor system. A noise model was proposed for output features of sensor and 4D EKF SLAM algorithm was executed via extracted features of scanner system. By this way, performance of 4D EKF SLAM algorithm is tested with real sensor data and output results are compared with 3D features. So in this work, contribution of using 4D features instead of 3D ones was examined via comparing performance of 3D and 4D algorithms with simulation results and real sensor data.

Localization and mapping are vital capabilities for a mobile robot. These two capabilities strongly depend on each other and simultaneously executing both of these operations is called SLAM (Simultaneous Localization and Mapping). SLAM problem requires the environment to be represented with an abstract mapping model. It is possible to construct a map from point cloud of environment via scanner sensor systems. On the other hand, extracting higher level of features from point clouds and using these extracted features as an input for mapping system is also a possible solution for SLAM. In this work, a 4D feature based EKF SLAM system is constructed and open form of equations of algorithm are presented. The algorithm is able to use center of mass and direction of features as input parameters and executes EKF SLAM via these parameters. Performance of 4D feature based EKF SLAM was examined and compared with 3D EKF SLAM via monte-carlo simulations. By this way / it is believed that, contribution of adding a direction vector to 3D features is investigated and illustrated via graphs of monte-carlo simulations. At the second part of the work, a scanner sensor system with IR distance finder is designed and constructed. An algorithm was presented to extract planar features from data collected by sensor system. A noise model was proposed for output features of sensor and 4D EKF SLAM algorithm was executed via extracted features of scanner system. By this way, performance of 4D EKF SLAM algorithm is tested with real sensor data and output results are compared with 3D features. So in this work, contribution of using 4D features instead of 3D ones was examined via comparing performance of 3D and 4D algorithms with simulation results and real sensor data.

Product development has both innovative and analytic sides. Starting from the requirements, a design suggestion is generated. In order to assess how well the envisioned design fulfils the requirements, it is sometimes necessary to build a computer model of it for the analysis. The overall motivation of the work presented is to reduce the time spent on creating the model by reusing knowledge gained from developing similar products by suggesting, building and evaluating IT-systems. To verify the systems real design examples, obtained from companies that have participated in the research projects have been used. The work is based on two major application examples. The first, involving the automated geometrical idealisation of die-cast parts (Paper I-III), and the second involving manufacturability of powder metallurgy pressed and sintered parts (Paper IV-VI). The work starts from the point in the product development process where it exists a design suggestion represented as an arbitrary format CAD-model. In the powder metallurgy case the object is to secure that the geometry is suitable for the production process. In the die-casting case the object is to automatically create an idealised version of the model for shell elements meshing. These two tasks have previously been treated as two separate cases, addressed by completely different software. This thesis suggests a common method for addressing the two cases. The method is based on converting the CAD-models, using the geometrical restrictions of the production processes, into a format with a specialised feature structure, parameterisation and construction history using a feature recognition approach. The features are then automatically reconstructed in a target CAD-system. The resulting, specialised CAD-model can be used for automated design and design evaluation purposes, demonstrated in the thesis. The models are therefore called DAR (Design Automation Ready)-models. The DAR-models are useful in that they separate the conversion from the subsequent treatment of the models providing modularisation, flexibility and user insight in the model structure. In that a construction history and parameterisation have be constructed in the target CAD-system, the advanced geometry manipulation and means for knowledge management often provided in modern CAD-systems can be accessed in a transparent and user manageable way. This extends the usefulness of the CAD-systems from involving only interactive work to managing all components sharing the same production process. / Teknologie Doktorsexamen

No description available.

No description available.

Developing process plans and part programs rapidly and correctly for CNC machine tools plays a vital role in manufacturing. This study is concerned with the development of a web-enabled virtual design and manufacturing application system for rotational parts. The object oriented methodology is used in the application development. Windows Distributed interNet Application (DNA) architecture which describes a framework of building software technologies in an integrated web and client-server model of computing, is employed in the system. The entire system was modeled with Unified Modeling Language (UML), which is an industry-standard visual modeling notation to express software development architectures. Feature-based design approach, being a practical way of linking the design with manufacturing, is implemented in the rotational part design. Users have to be registered in order to use the system. With the supplied web site, users can easily register to system. After registration, the user obtains a password and a unique username. Upon the user authentication, the user session starts. A typical user session involves new or past project selection, material selection, part sketching, blank size definition and cutting parameters determination. After all the steps are completed, process plan and part-program, which are required to manufacture the part on Denford Mirac CNC Turning Machine are generated and displayed on the web site and then saved to the database. The application enables the users to see past projects and to generate new process plan and part programs for different cutting parameters. The process plan, part program and the 2D wire frame drawing of the corresponding part are demonstrated on the web site for the chosen projects.

No description available.

No description available.

Visual attention improves sensory processing, as well as perceptual readout and behavior. Over the last decades, many proposals have been put forth to explain how attention affects visual neural processing. These include the modulation of neural firing rates and synchrony, neural tuning properties, and rhythmic, subthreshold activity. Despite the wealth of knowledge provided by previous studies, the way attention shapes interactions between cortical layers within and between visual sensory areas is only just emerging. To investigate this, we studied neural signals from macaque V1 and V4 visual areas, while monkeys performed a covert, feature-based spatial attention task. The data were simultaneously recorded from laminar electrodes disposed normal to cortical surface in both areas (16 contacts, 150 μm inter-contact spacing). Stimuli presentation was based on the overlap of the receptive fields (RFs) of V1 and V4. Channel depths alignment was referenced to laminar layer IV, based on spatial current source density and temporal latency analyses. Our analyses mainly focused on the study of Local Field Potential (LFP) signals, for which we applied local (bipolar) re-referencing offline. We investigated the effects of attention on LFP spectral power and laminar interactions between LFP signals at different depths, both at the local level within V1 and V4, and at the inter-areal level across V1 and V4. Inspired by current progress from literature, we were interested in the characterization of frequency-specific laminar interactions, which we investigated both in terms of rhythmic synchronization by computing spectral coherence, and in terms of directed causal influence, by computing Granger causalities (GCs). The spectral power of LFPs in different frequency bands showed relatively small differences along cortical depths both in V1 and in V4. However, we found attentional effects on LFP spectral power consistent with previous literature. For V1 LFPs, attention to stimuli in RF location mainly resulted in a shift of the low-gamma (∼30-50 Hz) spectral power peak towards (∼3-4 Hz) higher frequencies and increases in power for frequency bands above low-gamma peak frequencies, as well as decreases in power below these frequencies. For V4 LFPs, attention towards stimuli in RF locations caused a decrease in power for frequencies < 20 Hz and a broad band increase for frequencies > 20 Hz. Attention affected spectral coherence within V1 and within V4 layers in similar way as the spectral power modulation described above. Spectral coherence between V1 and V4 channel pairs was increased by attention mainly in the beta band (∼ 15-30 Hz) and the low-gamma range (∼ 30-50 Hz). Attention affected GC interactions in a layer and frequency dependent manner in complex ways, not always compliant with predictions made by the canonical models of laminar feed-forward and feed-back interactions. Within V1, attention increased feed-forward efficacy across almost all low-frequency bands (∼ 2-50 Hz). Within V4, attention mostly increased GCs in the low and high gamma frequency in a 'downwards' direction within the column, i.e. from supragranular to granular and to infragranular layers. Increases were also evident in an ‘upwards’ direction from granular to supragranular layers. For inter-areal GCs, the dominant changes were an increase in the gamma frequency range from V1 granular and infragranular layers to V4 supragranular and granular layers, as well as an increase from V4 supragranular layers to all V1 layers.