Category: Postgraduate

PGRs Research Presentations – November 2013

Amr Ahmed / / Postgraduate, Uncategorized

The monthly PGRs Research Presentations was held on Wed. 13th November, 2pm, Room MC0025 (Ground floor).

This session we had the following presentations:

Title: “Shaping human-aware navigation and human-robot joint motion using long-term adaptation“. Title:   “Data Analysis of Agent-based Crowd Simultion”

By: Christian Dondrup

By: Qinbing Fu
Abstract:Enabling a mobile service robot to move in a human populated environment is not only a question of safety but also of predictability, consistency, team work efficiency, and the general feeling of comfort of the human. This leads to a form of human-robot joint motion and human-aware navigation which is supposed to be most pleasant for the people involved. There are currently many approaches of solving this issue but most of them are built on constraints and static learning methods and not on long-term learning through interaction. The main focus of this thesis is therefore the creation of novel approaches to shape a robots spatial behaviour ”on-the-fly” using long-term experiences from engagements in joint movements with lay users and trying to find and understand adaptation needs and thereby create a predictable, readable and consistent robot behaviour.This first presentation will focus on state-of-the-art methods of social and human-aware navigation and a first study conducted to find gestures indicating adaptation needs to improve the feeling of comfort of human interaction partners and the likabilaty of the robot itself. Abstract: This work is based on clustering and visualizing an agent-based crowd simulation data in an airport, using K-means, Gaussian Mixture Model and Hidden Markov Model. With the help of these models, we mainly aim to verify the feasibility of our approach to analyse agents’ behaviour in the airport.

 

 

 

The meeting started by welcoming the new PGRs arrived.

After presentations and Q/A sessions, the meeting continued with:

* Congratulations Talal Al-Bacha, for his new baby girl.

* Brief information on the “Students Representative” duties/responsibilities/benefits.

* Election of the PGRs Students Representatives: 2 nominations and 2 positions available (based on a Rep for each 25 PGR).

* Students Reps for this academic year are: Christian Dondrup  and  Touseef Qureshi.

* Discussion about potential activities (social, trips,…etc.)

 

PGRs Research Presentations – October 2013

Amr Ahmed / / Postgraduate

The May’s PGRs Research Presentations was held on Wed. 9th October, 2pm, Meeting Room, MC3108 (3rd floor).

This session we had the following presentations:

Title: “Towards Verification and Validation of Crowd Simulation“. Title:   “COMPUTER VISION FOR THE FOOD INDUSTRY”

By: Oliver Szymanezyk

By: Eugenio Ivorra Martínez
Abstract:Human crowd motion is discussed to be driven by self-organising processes based on the local interaction amongst pedestrians. Despite past observations revealing the affect of social influences within pedestrian groups, they have been mostly neglected in the conceptualisation and implementation of simulated crowds. The impact of group structures on crowd dynamics is still relatively unknown, and developers are using assumptions and intuition within their models. In order to develop reliable and believable crowd simulation models that include group structures, it is necessary to use validated insights from interdisciplinary sociological and psychological research on crowds. Inspired by this need, we adapt computer vision technologies and crowd psychology research into a novel automated crowd simulation analysis tool. Our findings are implemented within an existing agent-based crowd simulation framework. The results demonstrate that the tool captures relevant quantitative data which returns insightful knowledge about the collective group movement of simulated pedestrians. Abstract: Computer vision is widely used for solve all kind of problems in the food industry (e.g. quality assurance of food products, checking the integrity of food packaging, and product manipulation).Until recently only 2D vision was used for these tasks. However, with the evolution of the technology, news fields have been opened for the industry in order to deal with unsolved problems or enhance the current solutions. Two of these fields include 3D vision and hyper-spectral imaging.

In this presentation I will introduce these fields and some solutions that we have developed. Our works are related mainly to quality assurance using non-destructive methods. This includes two works using hyperspectral imaging in which we have measured the food’s freshness, one for chicken breasts and another for smoked salmon.  In other work, we compared two different methods of capturing 3D images (time-of-flight and structured light) for scanning food. In addition, I will talk about the on-going and future work.

 

 

 

The Q/A was followed by a brief cath-up meeting, welcoming new PGRs arrivals and updates on the latest procedures.

 

PGRs Research Presentations – September 2013

Amr Ahmed / / Postgraduate

The May’s PGRs Research Presentations was held on Wed. 11th September, 2pm, Meeting Room, MC3108 (3rd floor).

This session we had the following presentations:

Title: “Simultaneous Localisation and Dense Mapping“. Title:   “Cue-based Aggregation with a Mobile Robot Swarm using a Novel Fuzzy-based Method”

By: Farhad Bazyari

By: Farshad Arvin
Abstract:Simultaneous Localisation and Dense Mapping.  Starting with introducing SLAM systems, then talking about dense mapping techniques and some of my work on this subject and autonomous exploration. Time permitting, I will very briefly mention ‘factor graphs’ as well. In addition to gaining massive popularity in SLAM community, factor graphs are extremely flexible structures that can be used in a wide field of research. So hopefully people whose research field are far from mine will also get something out of the talk. Abstract: In this work, we proposed a novel fuzzy-based method for cue-based aggregation with a mobile robot swarm. Our method is based on the state-of-the-art BEECLUST method that is inspired from thermotactic behavior of honeybee swarms. We assumed to have a sound source as the cue and designed a fuzzy-based controller with four microphone inputs and the output as the predicted direction of the sound source. We used 2 standard triangular membership functions, 12 fuzzy rules (3 for each microphone input) and centroid defuzzification method. We also proposed different variants of BEECLUST method, namely Naive and Vector-averaging methods. We compared the performance of our fuzzy-based method with BEECLUST and these methods using both Player/Stage simulator and custom-built autonomous robots (AMiR). Experiments were performed in two different settings: static and dynamic arenas.  In all the experiments, the fuzzy-based method performed better in terms of aggregation time.

 

 

 

The Q/A was followed by a brief catch-up meeting.

 

New Conference paper presented in the“ World Congress on Engineering 2013”

Saddam Bekhet / / Conference, Postgraduate, Publication

 Saddam Bekhet presented his accepted paper in “World Congress on Engineering 2013“.

The paper title is “Video Matching Using DC-image and Local Features ”

Abstract:

This paper presents a suggested framework for video matching based on local features extracted from the DC-image of MPEG compressed videos, without decompression. The relevant arguments and supporting evidences are discussed for developing video similarity techniques that works directly on compressed videos, without decompression, and especially utilising small size images. Two experiments are carried to support the above. The first is comparing between the DC-image and I-frame, in terms of matching performance and the corresponding computation complexity. The second experiment compares between using local features and global features in video matching, especially in the compressed domain and with the small size images. The results confirmed that the use of DC-image, despite its highly reduced size, is promising as it produces at least similar (if not better) matching precision, compared to the full I-frame. Also, using SIFT, as a local feature, outperforms precision of most of the standard global features. On the other hand, its computation complexity is relatively higher, but it is still within the real-time margin. There are also various optimisations that can be done to improve this computation complexity.

Well done and congratulations to Saddam Bekhet .

20130704_112542