Händelser: Studentarbete, Arkitektur, Bygg- och miljöteknik, Data- och informationsteknik, Energi och miljö, Kemi- och bioteknik, Matematiska vetenskaper, Material- och tillverkningsteknik, Mikroteknologi och nanovetenskap, Produkt- och produktionsutveckling, Rymd- och geovetenskap, Signaler och system, Sjöfart och marin teknik, Teknikens ekonomi och organisation, Fysik, Tillämpad IT, Tillämpad mekanik, Rymd-, geo- och miljövetenskap, Arkitektur och samhällsbyggnadsteknik, Bioteknik, Elektroteknik, Industri och materialvetenskap, Informations- och kommunikationsteknik, Mekanik och maritima vetenskaperhttp://www.chalmers.se/sv/om-chalmers/kalendariumAktuella händelser på Chalmers tekniska högskolaThu, 27 Jun 2019 14:12:19 +0200http://www.chalmers.se/sv/om-chalmers/kalendariumhttps://www.chalmers.se/sv/institutioner/math/kalendarium/Sidor/Masterarbete190821.aspxhttps://www.chalmers.se/sv/institutioner/math/kalendarium/Sidor/Masterarbete190821.aspxPresentation av masterarbete<p>MV:H12, Hörsalsvägen 1</p><p>​David Abraham Deniz: Multidimensional decision-making in early phase clinical trials</p>https://www.chalmers.se/sv/institutioner/e2/kalendarium/Sidor/Masterpresentation-Sachin-Madhusudhana.aspxhttps://www.chalmers.se/sv/institutioner/e2/kalendarium/Sidor/Masterpresentation-Sachin-Madhusudhana.aspxSachin Madhusudhana, MPSYS<p>Landahlsrummet (rum 7430), Hörsalsvägen 9, plan 7</p><p>​ Road profile classification</p><div>​Examinator: Balazs Kulcsar, Inst för elektroteknik</div> <div>Handledare: Anton Albinsson, Volvo Cars</div> <div><br /></div> <h2 class="chalmersElement-H2">Sammanfattning</h2> <div><br /></div> <div>The current active suspension systems are tuned extensively on different types of roads to have a good vehicle comfort and handling for safety and economic savings. Even though different modes can be chosen depending on the mode of the driver, such as comfort or dynamic, no adaptation of these modes are made for different road surfaces. The knowledge of the road profile estimation can be used to adapt the damping coefficient on active or semi-active suspension control systems to improve the ride comfort and handling of a car. Recent improvement in communication has enabled advancement in approaches towards vehicle safety which makes it possible for communication between the road infrastructure and the vehicle. This thesis investigates design of robust Hinf observer for road profile estimation and frequency analysis of the road irregularities. They are further classified into different road standards based on roughness spectra.<br /></div>https://www.chalmers.se/sv/institutioner/fysik/kalendarium/Sidor/Masterpresentation_Joakim_Berntsson_Adam_Tonderski_190902.aspxhttps://www.chalmers.se/sv/institutioner/fysik/kalendarium/Sidor/Masterpresentation_Joakim_Berntsson_Adam_Tonderski_190902.aspxMasterpresentation av Joakim Berntsson och Adam Tonderski<p>PJ, lecture hall, Fysikgården 2B, Fysik Origo</p><p>​ Titel på masterarbetet: Deep Learning for Sensor Failure Robust 3D Object Detection Masterprogram: Complex Adaptive Systems</p><h2 class="chalmersElement-H2">​Sammanfattning:</h2> <div><span style="background-color:initial">Autonomous Driving is the task of navigating a vehicle with reduced driver interaction. This requires accurate perception of the surroundings, which includes three dimensional object detection. In this area, deep learning methods show great results, and they usually use either images, point clouds or a fusion of both. These methods are often evaluated with full sensor availability. However, in a safety critical system, unexpected scenarios such as sensor failure must be accounted for. Therefore, the objective of this thesis is to develop a deep learning architecture that is robust against sensor failure.</span></div> <div> </div> <div>To accomplish this, we have designed an architecture that is heavily inspired by leading LIDAR object detection models. In order to be robust when the LIDAR is unavailable, we perform a learned projection from the 2D image to an artificial 3D point cloud. This is possible due to the fact that deep learning depth estimation models have reached a high performance. The two point clouds are merged into a common representation, which allows the model to perform the detections jointly and thus work if either sensor fails. The final contribution is a novel training procedure with varying sensor availability.</div> <div> </div> <div>The results show that the model is indeed robust against sensor failure. However, it is not quite on par with the respective state-of-the-art models that specialize on camera, LIDAR or fusion. We are confident that the results can improved considering the limited amount of tuning and multiple ideas that are left as future work. Finally, we show that sensor failure robustness can help the model generalize in a broader sense. </div>