Real Time Decision Making For Driverless City Vehicles.doc

Share Embed Donate


Short Description

Download Real Time Decision Making For Driverless City Vehicles.doc...

Description

REAL TIME DECISION MAKING FOR DRIVERLESS CITY VEHICLES

2. DEFINITIONS AND APPROACH 2.1 Solution Requie!ent"# De$i"ion M%&in' In(ut %n) Out(ut 2.1.1 2.1.1 De$i"ion De$i"ion M%&in' M%&in' Requie Requie!ent !ent"* "* The specification of requirements is a

fundamental part of the development cycle of any complex system, in order to enable its verification and validation . Regarding Regarding the developmen developmentt of the real-time decision decision making making module, module, the specification of requirements is crucial. Without a specification for this module, the impleme implementa ntatio tion, n, and eventu eventually ally human human lives lives ill ill depend depend on the softar softaree developers! interpretation of requirements, ithout the possibility to prove the functional correctness. " discussion on ho to develop a correct, unambiguous and complete specification for the decision making module for driverless city vehicles is beyond the scope of this ork. #oever, ithout the loss of generality, it can be assumed that such a decision making specification requires the decision making module to respond to certain traffic conditions in a defined ay, i.e. certain traffic conditions imply the execution of certain driving maneuvers. The occurrence of any traffic condition is modeled in this ork by a set of  discrete events. These discrete events are sent to the decision making module in

the form of World World $odel %vents. &epending &epending on a defined defined set of discrete events, the specification defines hich driving maneuvers can be performed safely. Table ' shos examples of such discrete events. " detailed and complete specification ould probably lead to a table containing thousands of such discrete events and driving maneuvers. (onsequently, the decision making module is modeled as an event-driven system )*+. #aving in mind the complexity of city traffic, a complete specification for a typical decision making module ill consist of a large number of defined inputsi.e. discrete events, and a large number of outputs i.e. driving maneuvers, hich are logically interrelated in a complex ay. TA+LE TA+LE I E,AMPLES OF DISCRETE EVENTS OF RELEVANCE.

REAL TIME DECISION MAKING FOR DRIVERLESS CITY VEHICLES

&iscrete %vent

%xplanation

/ stopped vehicle in front 0/ oncoming lane free */ pedestrian in front . . .

1topped vehicle as detected in front. The oncoming lane is obstacle free. 2edestrian crossing the road. . . .

2.1.2 In(ut 1*   The World $odel/ The World $odel represents the driverless

vehicle!s vie of its road environment. Therefore, it represents an input to the decision making module. The World $odel merges a priori given information e.g. road infrastructure ith traffic features hich are perceived during the vehicle!s movement e.g. dynamic obstacles. The data contained in the World $odel is constantly updated in real time and provided in to forms/ as World $odel %vents, and additionally in the form of an ob3ect-oriented structure.

" World $odel %vent is defined in this ork as a discrete event implemented as a 4oolean variable. " World $odel %vent represents the availability or state of  certain information generated by the World $odel. The purpose of World $odel %vent is to notify other softare modules about the state of certain predefined conditions,hich are relevant for the execution of  driving maneuvers. 1uch conditions can be, for instance, related to traffic situations e.g. 5pedestrian in front of the vehicle6, but might as ell be related to the availability of information e.g. 5721 locali8ation available6. The state of all defined World $odel %vents is provided as a k-tuple/ W$events 9,0, ...,k/ l ∈:true, false; here each element l l 9, 0, .., k represents a discrete event. The k-tuple W$events is updated cyclicly, and the World $odel notifies other  softare modules about the occurrence of certain conditions as defined for each event.

REAL TIME DECISION MAKING FOR DRIVERLESS CITY VEHICLES

'n addition to World $odel %vents, full access is provided to all data stored in the World $odel, through an ob3ect oriented data structure. This data structure contains more detailed information, such as exact distances to obstacles.#oever, a detailed discussion about the World $odel struc-ture is beyond the scope of this ork. 2.1.- In(ut 2*  The Route 2lanner/ 'n many situations, the planned route has a

significant impact on decision making.
View more...

Comments

Copyright ©2017 KUPDF Inc.
SUPPORT KUPDF