Search

This Deliverable 6.2 of the TransAID project presents and evaluates the simulation results obtained for the scenarios considered during the project's first and second iterations. To this end, driver- and AV-models designed in WP3, traffic management procedures developed in WP4, and V2X communication protocols and models from WP5 were implemented within the iTETRIS simulation framework. Previous main results from Deliverable 4.2, where baseline and traffic management measures without V2X communication were compared, have been confirmed. While not all TransAID scenarios' traffic KPIs were affected, the realistic simulation of V2X communication has shown a discernible impact on some of them, which makes it an indispensable modelling aspect for a realistic performance evaluation of V2X traffic scenarios. Flaws of the first iteration's traffic management algorithms concerning wireless V2X communication and the accompanying possibility of packet loss were identified and have been addressed during the project's second iteration. Finally, lessons learned while working on these simulation results and assessments have additionally been described in the form of recommendations for the real-world prototype to be developed in WP7. We conclude that all results obtained for all scenarios when employing ideal communication confirmed the statistical trends of the results from the original TM scenarios as reported in Deliverable 4.2 where no V2X communication was considered. Furthermore, the performance evaluation of the considered scenarios and parameter combinations has shown the following, which held true in both the first and second iterations: (1) The realistic simulation of V2X communication has an impact on traffic scenarios, which makes them indispensable for a realistic performance evaluation of V2X traffic scenarios. (2) Traffic management algorithms need to account for sporadic packet loss of various message types in some way. (3) Although important, the realistic modelling and simulation of V2X communication also induces a significant computational overhead. Thus, from a general perspective, a trade-off between computation time and degree of realism should be considered.

The TransAID project defines, develops and evaluates traffic management measures based on V2X equipped road infrastructure, primarily via simulations,to eliminate or mitigate the negative effects of Transition of Control (ToC) along Transition Areas in future mixed traffic scenarios where automated, cooperative, and conventional vehicles will coexist.This document aggregates, integrates, and analyses the results of the TransAID work packages. For each aspect of TransAID the major findings are presented and discussed.As a basis for the simulation studies several vehicle models were implemented successfully to create the right behaviour for lane changing(including cooperative versions), car following (including (C)ACC)and ToC/MRM algorithms. These models were created using a solid theoretical background, however, the availability of real-world data for input and calibration was very limited. From the baseline simulation runs we found that ToCs do not significantly disrupt traffic flow performance unless CAVs establish increased car-following headways during the ToC preparation phase. Disruptions escalate in case of CACC driving, increased share of CAVs in the fleet mix, and the occurrence of multiple ToCs within a narrow temporal window and spatial domain. Furthermore, in the case that a ToC is unsuccessful or not possible, unmanaged MRMs (taking place in lane and not being guided towards safe spots) can induce significant traffic disruption as well. On the other hand, simulation results indicated that cooperative lane changes minimize the frequency of ToC/MRM and their consequent adverse impacts on trafficflow operations. The benefits of cooperative lane changing are amplified with increasing share of CAVs and especially upstream of lanedrop locations.Building upon the vehicle models, simulations and the defined use cases, specific traffic measureswere developed to mitigate the effects of ToC events in transition areas. The traffic measures were implemented to study their effectiveness. Specifically, for each of the selected use cases the effects of the TransAID measures are evaluated regarding emissions, safety and efficiency. There is a trade-off between traffic safety versus traffic efficiency (as measured via throughput and travel times). It is often inherently difficult or even impossible to optimise both in the same context. Hence, typically a policy choice needs to be made, as to which of the two will have to be prioritised. Otherwise, results either improved or remained similar for all use cases and KPIs, with the exception of use case 3.1 (see Section 2.2.2 for details and Table 1at the end of Chapter 6). All use cases have in common that a reduction of MRMs is possible by providing infrastructure advice. Such advice, and the availability of safe spots, clearly reduces the number of stopped vehicles blocking the road.There is also a heavy dependence of the results on the mixture of vehicle types, in addition to the observation that less efficient traffic management performance is obtained for a higher LOS. The latter is in part logical, as for higher LOS there is more prominent congestion and the physical limits of the infrastructure remain a hard obstacle. By itself this is not a problem for TransAID, as the focus of the traffic management schemes is to prevent/postpone traffic breakdowns before they occur. While implementing and testing the traffic measures TransAIDalso identified or created the needed message sets and protocols to implement the measures using V2X communications. To that end, no new message sets wereneeded, but (minor) extensions to CAM, DENM, MCM and MAPEM were necessary. Especially MCM from the Manoeuvre Coordination Service (MCS) is key to multiple types of use case. Therefore, it is necessary to define a MCS that is valid for all types of scenarios. Aligned with the work of ETSI and by actively contributing, TransAID has proposed a MCS where the infrastructure takes an active role to facilitate the manoeuvres of vehicles and to increase the overall traffic flow and safety. The traffic management measures designed in TransAID also require that CAVs and road infrastructure units have an accurate perception of the environment. In addition to the MCS, TransAID has contributed to the evaluation and evolution of ETSI's Collective Perception Service (CPS) for cooperative perception. We have demonstrated that cooperative perception can improve CAVs perception capabilities when the trade-off between the perception capabilities and communications performance is balanced. Furthermore, the reliability of V2X communications has been addressed in TransAID using different and complementary techniques: compression, congestion control and acknowledgements.Besides the V2X communication, the communication to unequipped vehicles was of importanceand consisted of two parts. On the one hand, infrastructure needs to inform unequipped vehicles about issues on the road. On the other, automated vehicles themselves should provide information about their actual state to their surroundings, to avoid negative impacts.With regards to the infrastructure information, it needs to be mentioned that visual information on signs, variable or static, will never be as precise as V2X communication could be, esp. when looking to individual advices. Nevertheless, infrastructure can provide valuable information also to unequipped vehicles by signage, e.g., in terms of speed limits, distance (gap) advice or dynamic lane assignments.It will be required to create additional road signs dealing with automated vehicles, at least showing that, e.g.,an area is prohibited for automated vehicles or an area where only automated vehicles are allowed.Regarding signals from automated vehicles, TransAID's solution of having LED light strips at the back of AVswill be beneficial in any case, but the exact content of such lights needs to be defined by performing more detailed analyses of such components. This goes to all external and dynamic HMI components of automated vehicles. In this light, it will be crucial to have an intuitive way of understanding the automation related additional information. One key question in this area is if driving with enabled automation should be indicated by an additional external light, and if so, where should this light be and whatcolour?Combining the work on the traffic measures and communications, the iTETRIS framework was used to evaluate the selected use cases while deploying the traffic measures using V2X. The goal was to see if the V2X communications impacted the effectiveness of the measures in any way.After adding V2X, the simulation results forthe project's first and second iteration use cases showed very similar results to the previous evaluation. All traffic measures were found robust enough to show the same results as with idealV2X, even in light of increased traffic demand and thus more V2X enabled vehicles.There were some minor differences between the realistic V2X and ideal V2X implementations, but those could be traced back to easily fixed technical aspects (see Section 2.4 for details). As a final step in our use case assessment,the feasibility of measures and communications introduced were implemented in real-world demonstrators. The real-world implementation was done by performing three different feasibility assessments. Two of them have been performed on test tracks in Germany, and one on public roads in The Netherlands. On the test tracks, several detailed tests of all scenarios have been performed, revealing that all traffic management measures could be successfully integratedand applied to automated vehicles in all use cases and scenarios. This includes the successful setup of the RSI and the automated vehicles. It has to be mentioned, though, that the implementation was done in a prototypic way.The development of related series products would require much more testing under real world conditions, which will be challenging at the current time since no highly automated vehicles are present on the roads. Nevertheless, it is very important to start the investigations at present times. As already described in Section 3.3, standardisation of messages is happening already now, and it was very important to include the role of the infrastructure at this stage. The detailed results of the real-world implementations per use case can be found in Section 2.5.In addition to the design and technical implementation of traffic measures in simulation and the real-world, TransAID gained some insights on issues of a less technical nature. For example, it was determined a close collaboration between OEMs and (N)RAs would be beneficial in the identification and managing of TAs. To facilitate such a collaboration TransAID proposes a traffic management frameworkin the form of an intermediary service provider, acting as a trusted (and possibly mandated) third party. The framework allows TransAID to be scaled up and generalised. We approached this from both a technical and a business-oriented perspective. For TransAID to become part of a complete traffic management system, we focused on the technical side on how to detect transition areas, select (and possibly combine) services, and then detect when they are most appropriately timed for deployment. To this end, detection can be done via the infrastructure (e.g. road sensors or even digital communication infrastructure), via the OEMs, or by comparing an infrastructure's newly-defined ISAD level (Infrastructure Support levels for Automated Driving; see Section 1.4.2) to theoperational design domain (ODD, see Section 1.4.1) of the vehicle.Considering the mentioned technical challenges (detecting TAs, selecting services, and timing their deployment), the intermediary service bridges all these parties in such a way that the detection of TAs is performed in a centralised way, and OEMs and (national) road authorities ((N)RAs) have a single point of contact for providing and receiving information about TAs.Another point where OEMs and (national) authorities could collaborate,is the legislation related to automated driving since an important gap in current modelling and legalisation is how (C)AVs would/should react when (given) advice and/or actions conflict with traffic laws. With the real-time coordinated instructions of a TMC, (C)AVs should drive adequately during their journeys. However, it is necessary to concern to what extent such instructions should/can be made, especially when considering legal issues.In addition, legal aspects like the definition of special signage for automated vehicles and their handlingalso need to be considered, as those aspects will take time. This also means signage at the roadside, including VMS content, and signage from automated vehicles to surrounding traffic.Collaboration is also required regarding the definition and standardisation of V2X messages and protocols. The mechanisms proposed in TransAID to improve the reliability of V2X messages can be key in the near future. In general, V2X communications solutions require to be incorporated into standards to be effectively deployed. That is the case for, for example, collective perception solutions, message generation rules for manoeuvre coordination, V2X message compression or broadcast acknowledgement mechanisms. In TransAID we have been intensively working to promote and disseminate all the proposed solutions in top-tier journals and international conferences, as well as in organisations like ETSI and C2C-CC. The above shows a broad range of aspects studied by TransAID in the very dynamic and rapidly evolving field of automated driving.To provide links to additional information and to place the work of TransAID into context, Chapter 5 provides an overview or close related initiatives.

In light of the increasing trend towards vehicle connectivity and automation, there will be areas and situations on the roads where high automation can be granted, and others where it is not allowed or not possible. These are termed ‘Transition Areas’. Without proper traffic management, such areas may lead to vehicles issuing take-over requests (TORs), which in turn can trigger transitions of control (ToCs), or even minimum-risk manoeuvres (MRMs). In this respect, the TransAID Horizon 2020 project develops and demonstrates traffic management procedures and protocols to enable smooth coexistence of automated, connected, andconventional vehicles, with the goal of avoiding ToCs and MRMs, or at least postponing/accommodating them. Our simulations confirmed that proper traffic management, taking the traffic mix into account, can prevent drops in traffic efficiency, which in turn leads to a more performant, safer, and cleaner traffic system, when taking the capabilities of connected and autonomous vehicles into account.

The TransAID project defines, develops and evaluates traffic management measures based on C-ITS equipped road infrastructure to eliminate or mitigate the negative effects of Transition of Control (ToC) along Transition Areas in future mixed traffic scenarios where automated, cooperative, and conventional vehicles will coexist. This document summarises the results of the stakeholder consultation activities of the TransAID project. For TransAID, the most relevant stakeholders are transport authorities and policy makers, road operators, vehicle manufacturers and suppliers, road infrastructure and traffic service providers, academia and knowledge institutes, future product owners and standardisation bodies. The consultation activities aimed to gather feedbacks on the project choices (selected use cases, scenarios, modelling solutions, implementation approaches) as well as on the achieved results. Such feedbacks were necessary to confirm the validity of the project's work, and to adapt, whenever possible and needed, its implementation while running. Consultation also allowed obtaining the stakeholders' view on the impact of prospective automated vehicles introduction. Finally, the stakeholders were asked about their ambitions and interests related to role and responsibilities in future scenarios of automated vehicle presence. Hearing about these last two aspects was necessary to identify possible activities beyond the TransAID project's duration. A summary of 9 stakeholder consultation events is provided in this deliverable: Main stakeholder workshops - TransAID-MAVEN-CoExist Stakeholder workshop, 10 October 2017, Brussels - TransAID-INFRAMIX stakeholder workshop, 9 October 2019, Graz - TransAID final event, stakeholder workshop, 2 July 2020, online International liaison activities - TransAID-U.S. CAMP expert meeting, 25 July 2019, Detroit - TransAID + ITS Japan / UtmobI expert meeting, 7-8 April 2020, online Additional stakeholder consultation opportunities - TransAID session and survey, 8 June 2019, IEEE-IV, Paris - EU EIP workshop on ODD, 1 October 2019, Turin - International workshop on ODD, 22 October 2019, Singapore - Joint dissemination of H2020, CEDR projects and other initiatives related to CAVs and Infrastructure, 3 March 2020, Brussels For each stakeholder consultation event, this deliverable gives a detailed description of the scope and aim, participants, plenary and break-out sessions, survey results (when applicable) and implications to the TransAID work. The main common findings that have been identified throughout the various events are listed in the following along with the major implications for TransAID and/or any similar follow-up activity: - Managing mixed traffic in transition areas is still an almost unexplored field. Despite transition areas are recognized as a prospective problem, very little is known or has been studied about it. This simple acknowledgment confirms the need and timeliness of TransAID. - Due to the uncertainty about many technical aspects related to the introduction of AD, it was difficult for transport authorities and traffic managers to provide insights on aspects to be considered for the selection of TransAID use cases and scenarios. Nevertheless, the majority of use cases and scenarios finally selected by TransAID were recognized to be reasonably generic, yet addressing recurrent problems, and hence deemed useful for real-world application. - Connectivity was recognized as a key enabler to extend the Operational Design Domain (ODD) of AD. In this sense, most experts foresee application of hybrid solutions with both ITS-G5 and cellular connectivity capabilities. For this purpose, TransAID's assumption of digitalizing the road infrastructure with additional sensing, computing and communication capabilities was acknowledged to be correct, even if maybe not realisable in the short term. - TransAID traffic management allowing the road infrastructure to provide additional information to CAVs was also recognized as a valid approach in almost all the consultation and twinning events. In this context, European and Japanese stakeholders firmly defend the use of infrastructure support for automated driving and even highlight the need to adapt traffic rules for automation or change the legal frameworks (e.g. authorize the road infrastructure to provide advices that break the traffic rules if needed). On the contrary, US stakeholders are very hesitant and fear possible financial consequences resulting from liability issues. - Remote operation is an emerging possible solution at least for management of level 4 automated public transport (e.g. autonomous shuttles or pods) in edge cases and transition areas, when the vehicles operate without a steward in the vehicle. This use case is not considered by TransAID, and hence it would be interesting to investigate its effectiveness with similar evaluations means as those utilized by TransAID. - Sensitivity of information around OEMs' current and future implementations has prevented obtaining commonly recognized functional descriptions reusable for modelling and simulations of automated vehicles behaviours. Despite that, the modelling solutions developed in the project under the consultancy of Hyundai were considered adequate and meaningful by the inquired stakeholders. - For the TransAID measures to work effectively, vehicle AD capabilities shall be known by the infrastructure, and infrastructure (support) capabilities shall be known by OEMs. Consequently, sharing this data in both directions is needed. From the consultation, this approach was welcomed by infrastructure stakeholders while OEMs were more hesitant mostly due to competition and liability implications arising from sensitivity of the shared information. To preserve sensitivity of information, the TransAID "intermediary service" concept (see TransAID D4.3) was generally supported, but its practical application in real-world deployment scenarios would need to be proven. - In addition, it is considered needed to derive clear and unambiguous definitions of AD ODDs for adoption at both OEMs and infrastructure side. Transport Authorities could use this information for allowing vehicles of different automation capabilities to use specific roads or to provide additional physical of digital support where needed. Nevertheless, defining ODDs is a complex task for the involved stakeholders, and despite initial activities have been started and proposals have been made, there is still a long way to go. - From an even more generic perspective, transport authorities (especially cities) are mostly interested in fulfilling their greener, safer and comfort goals and see AD as an opportunity in that direction. For the moment being, they are not favouring a particular automated transport mode in a proactive way. Rather, they seem to be monitoring the situation to apply reactive policies when AD introduction will be more mature and clearer. In fact, adoption of alternative policies like strategically "managing" private CAVs vs. fostering use of MaaS with public automated shuttles will depend on penetration. For TransAID, it is irrelevant which way authorities will choose, as the proposed TransAID solutions can apply irrespectively of the selected automated transport mode. As it can be seen, some of the findings from the consultation events reflect the uncertainty associated to vehicle automation and its evolution in the coming decades. Nevertheless, from the sequence of stakeholder consultation events a steady progression in the collective understanding of the relation between vehicle automation and infrastructure could be observed. The possible implications to the stakeholders involved became also clearer at subsequent events.

TransAID develops and demonstrates traffic management procedures and protocols to enable smooth coexistence of automated, connected, and conventional vehicles, especially at Transition Areas. A hierarchical approach is followed where control actions are implemented at different layers including centralised traffic management, infrastructure, and vehicles. This document summarises the results of the stakeholder consultation activities of the TransAID project. In the context of TransAID, the most relevant stakeholders are authorities and policy makers, road operators, vehicle manufacturers and suppliers, road infrastructure and traffic service providers, test and certification institutes, academia and knowledge institutes, future product owners and standardisation bodies. The consultation activities aimed to gather feedbacks on the project results, as well as to hear the stakeholders' view on the impact of prospective automated vehicles introduction. Most importantly, the stakeholders were asked about their ambitions and interests related to role and responsibilities in future scenarios of automated vehicle presence A summary of 7 stakeholder consultation events is provided in this deliverable: - TransAID-MAVEN-CoExist Stakeholder workshop, 10 October 2017, Brussels - TransAID-MAVEN-CoExist-INFRAMIX Expert meeting, 23 October 2018, Greenwich - TransAID session and survey, 8 June 2019, IEEE-IV, Paris - TransAID-U.S. CAMP expert meeting, 25 July 2019, Detroit - EU EIP workshop on ODD, 1 October 2019, Turin - TransAID-INFRAMIX stakeholder workshop, 9 October 2019, Graz - International workshop on ODD, 22 October 2019, Singapore For each stakeholder consultation event a description is given of the scope and aim, participants, plenary and break-out sessions, survey results (when applicable) and implications to the TransAID work. What can be observed from the sequence of stakeholder consultation events is that there is steady progression in the collective understanding of the relation between vehicle automation and infrastructure and the possible implications to the stakeholders involved. By now it seems that there is a common interest, also by vehicle manufacturers, to develop a comprehensive standard and/or taxonomy for classifying operational design domains (ODDs) of automated vehicle systems. The main findings from these events underline the uncertainty associated with the state-of-the-art of vehicle automation and its evolution in the coming decades. From an innovation standpoint these are exciting times, but as we have experienced, the uncertainties will not disappear soon or new uncertainties will arise. Moreover, since Cooperative, Connected and Automated Mobility and Digital and Physical Infrastructure are such new areas of innovation, the stakeholder consultation did not provide all the answers while for many subjects, nobody has the answer yet.

With advances in technology, human-machine interfaces have become commonplace. Their design requires a great deal of engineering efforts to make them functional and accessible. One of these engineering efforts is the embedding of voice control. This improves the accessibility for people with a physical disability. In common speech-enabled command-and-control applications, the spoken commands are restricted to a predefined list of phrases and grammars. These conventions work well as long as the system does not have to stray too far from the conditions considered by the designer or from the characteristics of the training material. Speech technology would benefit from training during usage; learning the specific vocalizations and the emerging expressions of the user. Designing a vocal user interface (VUI) model from this developmental perspective would widen accessibilitynbsp;caternbsp;users with non-standard or dysarthric speech. The research in this dissertation is aimed at the development of a self-taught VUI that learns speech commands from the user while it is operational. Tonbsp;end, we adopt and introduce different procedures in order to build a VUI-model that learnsnbsp;a few learning examples. A learning example consists of two sources of information: the spoken command and the demonstration of the commanded action.nbsp;sources of information are converted to fixed-length utterance-based vectors. The followed approach links the acoustic patterns that are embedded in the spoken utterances to the concepts that jointly define the meaning of the utterance. The method represents the data by its recurrent acoustic and semantic patterns and the incidence of these patterns in the data. Since these patterns are embedded in the data, the representation of the data has a significant influence over the performance of thenbsp;model. A thorough analysis ofnbsp;representations resorting to speaker-dependent and speaker-independent data resources, is made. Attention is also given to the representation of thenbsp;action. The representation of the commanded action consists of an incidence vector representing the semantic content of the demanded action. Users are non-experts innbsp;a VUI, therefore, errors such as uttering an incomplete command or pushing a wrong button,nbsp;emerge. We demonstrate robustness against these kinds of errors. Another issue pertaining to semantics is the correlation between relevant concepts in a spoken utterance. This dependency is an additional source of information. We exploit this information and compare different semantic structures pertaining to these semantic dependencies.nbsp;nbsp; With the focus on the learning process rather than on the resulting model, we develop procedures for incremental and adaptive learning. By exploiting a semi-Bayesian procedure called maximum a posteriori (MAP) estimation, the VUI model can be made to learn incrementally, one utterance at a time. Incremental learning procedures are developed at the level of the basic acoustic atoms and at the level of the word models. They are compared with their batch learning variants and yield comparable accuracy. The implementation of a forgetting factor makesnbsp;models adaptive to changes in the speech of the user. The learning curves are an assessment of the quality of learning in function of the amount of training data. We analyse the learning curves for all these developments by numerous experiments in realistic learning scenarios implemented on computer. By this, we acquire a sense of the systemnbsp;performance in a real-world training environment. Thenbsp;of the VUI in its operational context and the training of the VUI by the user are the twonbsp;important key aspects that inspired the conception, the developments and the research questions in this study. Ons B., ''The self-taught speech interface'', Proefschrift voorgedragen tot het behalen van het doctoraat in de ingenieurswetenschappen, KU Leuven, May 2015, Leuven, Belgium. status: published

Tessema N., Ons B., van de Loo J., Gemmeke J.F., De Pauw G., Daelemans W., Van hamme H., ''Metadata for Corpora PATCOR and Domotica-2'', Technical report KUL/ESAT/PSI/1303, KU Leuven, ESAT, July 2013, Leuven, Belgium. ispartof: Technical report KUL/ESAT/PSI/1303, KU Leuven, ESAT, Leuven, Belgium status: published

ispartof: Kragujevac Journal of Mathematics vol:35 issue:2 pages:317-325 status: published

ispartof: Kragujevac Journal of Mathematics vol:33 pages:5-15 status: published

Plateau’s irradiation phenomenon in particular describes what one sees when observing a brighter object on a darker background and a physically congruent darker object on a brighter background: the brighter object is seen as being larger. This phenomenon occurs in many optical visual illusions and it involves some fundamental aspects of human vision. We present a general geometrical model of human visual sensation and perception, hereby taking into account the law of Fechner in addition to the anisotropic smoothing that was introduced in [1], and explicitly illustrate its meaning for irradiation illusions of Helmholtz and Kitaoka. ispartof: Kragujevac Journal of Mathematics vol:34 pages:5-24 status: published

This doctoral project tackles a number of specific research issues regarding the perception of contour shape. Shape perception starts from the light excitation of the 2D-retinal surface, and contours in the retinal image reveal probably the most important information for representing the shape of objects. Many significant views on object representations incorporate the importance of contour information in object recognition (Biederman, 1987; Edelman, 1998, 1999; Graf, 2002; Marr & Nishihara, 1978). More often than not, it is required to make sense of retinal images from different viewpoint angles and from objects that are slightly different from any object that has been encountered before. On the one hand, the visual system should try to ignore shape properties that do not affect the meaning of an object; on the other hand, it should track all shape properties that are able to change the identity of an object. In this dissertation, we will try to solve these contradictory requirements by investigating image transformations that do and do not affect the identity of an object (see also Graf, 2002). In Part 1, we advocate that shape properties that are diagnostic for the identity of objects at the basic-level of category inclusiveness are affected by most topological transformations, but not metric and affine transformations. In "1.1 Linear and curvilinear spatial transformations", we call topological transformations planar curvilinear shape changes and affine transformations planar linear shape changes and we explain how they incorporate many significant shape features of different theoretical views as special cases. in "1.2 Experiment 1", we present the results of a delayed match-to-sample task that we administered to children of the age of 3 to 7 years and we found that children became gradually more sensitive to the curvilinear shape changes that are relevant for between-category distinctions. Moreover, we found that the sensitivity to both kinds of shape differences becomes more uncorrelated with age. In "1.3 Experiment 2", we present an experiment in which we added nonsense words to each individual object and we investigated how children formed stimulus-word associations. First, we were able to replicate the results of the first experiment. Moreover, we found that the confusion in naming reflects confusion in shape perception. We also found evidence for an early sensitive period in development for acquiring word-stimulus associations. In Part 2, a particular example of a linear (aspect ratio) and a curvilinear (curvature) transformation was explored in more depth in adults. First, a method was introduced to measure the participant's interpretation of both transformational dimensions (see "2.1 Measuring the preferred perceptual basis in mental shape space"). Second, the independent processing of the dimensions derived by the method was explored in various tasks (see" 2.2 Experiment 3" for speeded categorization, and see "2.3 Experiment 4" for shape adjustments). The results show that the derived dimensions were fairly separable and that the adjustments on both dimensions did not correlate with each other. Similar to the older age group of part 1, adults seemed to process the linear (aspect ratio) and the curvilinear (curvature) shape differences independently. In Part 3, it is explained how the bias in an adjustment task can serve as an indicator of category formation. In "3.1 Experimental proposal 1", we investigate whether a visual category precursor has an influence on the adjustments of a contour dimension when the category precursor and the dimension are integral or separable (the precursor and the dimension are always related to aspect ratio). The integration of both dimensions creates the typical categorical biases in adjustments that have been observed before (see Sailor & Antoine, 2005; and also, Huttenlocher & Hedges, 2007) . In "3.2 Experimental proposal 2", we report a bias that we measured in a pilot adjustment task concerning the orientation of dot patterns. We tend to fit the bias into the model of spatial categories by Huttenlocher, Hedges and Duncan (1991). In their model, the angular bias is explained by prior distributions of angular categories. However, those categories are problematic in the transformational view of contour perception. Therefore, in "3.3 Experimental proposal 3", we intend to investigate the bias for more sophisticated contours consisting of multiple parts and more curvature segments and we tend to infer which reference system the visual system uses for contour coding. The inferred reference system can provide new insights into the transformational approach. status: published

A model is proposed that elegantly unifies the traditional exemplar and prototype models. These two models are extreme cases of the proposed varying abstraction model. The unifying model further makes room for many new intermediate pseudo-exemplar models. A preliminary analysis using Medin and Schaffer's (1978) 5-4 structure pointed to such an intermediate model that outperformed the prototype and exemplar models. 11 ispartof: pages:2277-2282 ispartof: Proceedings of the Annual Conference of the Cognitive Science Society vol:27 pages:2277-2282 ispartof: Annual Conference of the Cognitive Science Society, Stresa, Italy location:Stresa, Italy date:21 Jul - 23 Jul 2005 status: published