Virtual Race Strategist
Cato is based entirely on the technology of Jona, the Virtual Race Engineer. Therefore, all concepts from the associated documentation, especially for installation and configuration, also apply here.
As a complement to Jona, your Virtual Race Engineer, this new Assistant will accompany you during your races as a Virtual Strategist. Cato will have complete knowledge over the race session incl. lap times of you and your opponents, the current race positions, pitstop events, weather changes, and so on. You can request updates for all these informations by asking Cato for the position, the gaps to the car in front and behind and also request information about the current lap times of your opponents and wether they are closing in. All this, although very useful, if you are racing using VR, is only a part of Catos capabilties. Cato is furthermore able to develop appropriate pitstop strategies if you are stuck in traffic and he will be able to react to unforeseen events such as sudden weather changes and severe damage - all hand in hand with Jona, the Virtual Race Engineer. Cato currently supports Assetto Corsa, Assetto Corsa Competizione, rFactor 2, Le Mans Ultimate, RaceRoom Racing Experience, iRacing, Automobilista 2 and Project CARS 2.
Before we dig deeper into the inner workings, here is a typical dialog based interaction, to give you an understanding of the current capabilities of Cato.
Important: If you have multiple dialog partners active, for example Jona and Cato, it will benecessary to use the activation phrase the first time you start talking to a specific dialog partner.
(You are on the track, for example in the third lap. A lot of traffic has developed around you, and you want get more information about the current standings. You can call Cato anytime using one of the key phrases - see the section about phrase grammars for more information.)
Driver: "What is my position?"
Cato: "You are currently on P 12."
(This information might be a little bit outdated, since many of the simulation games update the position information only once per sector or even only once per lap.)
Driver: "What is the gap to the car behind me?"
Cato: "The gap currently is 1.1 seconds."
(You have seen in the mirror, that the car behind you closed in quite rapidly. Therefore you want to compare the last lap times.)
Driver: "Can you give me the lap times?"
(Cato will now tell you the lap times and the lap deltas for the car in front of you, the car behind and the leading car.)
Cato: "Your last lap time was 122.2 seconds. The car behind you is running a 121.3 and is therefore 0.8 seconds faster than you. 124.7 seconds is the time of the car in front of you. He is 2.5 seconds slower than you. The leader is running a 120.9 and is therefore 1.3 seconds faster than you."
(You regain your focus and try to overtake P 11. It will be a nice safety barrier between you and the faster car behind you...)
(Finally, after you've managed to gain P 11 and settled your pace, you want your strategist to run a simulation, how the race might develop in the next few laps.)
Driver: "What will be my position in 4 laps?"
Cato: "Understood. Please give me a few seconds."
(Cato runs a simulation in the background to calculate the possible race positions for the requested lap. Cato will use the average pace of all drivers and takes into account, how many seconds an overtake might need depending on the settings.)
Cato: "The simulation shows you on P 8."
(That is a very good message. You put your foot down on the pedal and push a little bit more to bridge the gap to the car in front of you.)
(After a few laps, you finally managed to gain P 8. Now it is time to plan for the upcoming pitstop. Maybe Cato can give you some recommendation...)
Driver: "What is the best lap for the next pitstop?"
Cato: "Understood. Please give me a few seconds."
(Cato again runs a complex simulation, taking into account your remaining fuel, the current race positions and the weather outlook. Cato will recommend the lap for the pitstop, where you will have the least traffic after the stop, or the lap where an undercut might be possible, taking into account the best lap for a tyre compound change, if applicable.)
Cato: "I recommend a pitstop in lap 17. Should I inform your Race Engineer?"
Driver: "Yes, please."
(Cato will hand over the information to your Virtual Race Strategist, which will handle the technical stuff and plan the pitstop.)
To have an error free session like this one, you must have a perfect setup for voice recognition. I strongly recommend using a headset, since alien noises might lead to false positives in Catos voice recognition. Please see the section on troubleshooting in the documentation for Jona, if you need some hints.
The installation procedure for Cato is the same as the installation procedure for Jona, which means, that you are also well prepared to use Cato, if you have everything setup correctly for Jona.
The same principles as described for Jona apply here as well, since Cato is based on the same technology as Jona.
I strongly recommed to memorize the phrases in the language you use to interact with Cato. You will always find the current version of the grammar files as actually used by the software in the Resources\Grammars folder of the Simulator Controller distribution. Or you can take a look at the files in the Resources\Grammars directory on GitHub, for example the German version Race Strategist.grammars.de.
Beside the builtin pattern-based voice recognition and the speech capabilities based on predefined phrases as described above, it is optionally possible to connect Cato to a GPT service like OpenAI or a locally hosted LLM, to dramatacilly improve the quality in conversation with the Assistant. And you can also extend the knowledge and reasoning capabilities. When the Conversation Booster or the Reasoning booster are configured (see here for more information about the necessary configuration steps), the full knowledge about the car state will be supllied to the LLM. In detail, this includes tyre pressures, tyre temeperatures, tyre wear, brake temperatures, brake wear, fuel level, fuel consumption, damage, and so on. When a pitstop is planned, the plan is available and the pitstop history is also available.
Cato can give you information about the weather development and might recommend a strategy change. You may disable this warning by using a special voice command:
[Please] No more *warning* [please]
As you might expect, the word "please" is optional. Only one option for warning is available at the moment: "weather warnings". After you have disabled the warning (it is enabled by default), you can reenable it with the following command:
[Please] Give me *warning* [please]
As an alternative, you can disable unwanted talking completely by saying:
Be quiet please
To reactivate the Assistant use:
I can listen now
These commands are also available as "Mute" and "Unmute" plugin actions, which can be configured for a Button Box or a Stream Deck, for example.
Good to know: Even if Cato is muted, you can interact with the Strategist, for example to ask for a undercut simulation. If a chain of conversation is initiated by you, the driver, Cato will react normally. The Strategist will also communicate normally with regards to all strategy related topics, if a strategy is active for the current race.
Cato will be active during practice and race sessions by default, although the Assistant will be of not much help in a practice session, since it only collects data for future race strategy development purposes. You can configure the sessions, where Cato collects telemetry data using the settings in the "Session Database".
Cato will always be active during a race, even if you have disabled data collection for races. Cato ca support you in a couple of different areas while you are running a race.
-
You can activate the Assitant anytime using the activation phrase and ask then for information about current lap times, current and possible future standings and so on. This is also possible using controller actions in most cases.
-
Cato can simulate future race situations based on the knowledge about your driving and all the other participants. This includes the devlopment of the standings as well as a recommendation for the best possible lap for an upcoming pitstop.
-
And Cato is able to guide you through the race using a strategy that was prepared before the race using the "Strategy Workbench" tool. This includes the announcement of an upcoming pitstop, as well as a cooperation with the Race Engineer to prepare and handle the pitstop.
Normally, Cato will not contact you on its own as often as Jona does. Most of the time, you must ask Cato specifically for its support. An exception is, when you have an active race strategy or when a weather change will require you to conduct an unplanned pitstop for tyre change.
Important: In the default configuration, Cato will be activated in the first lap of a session. This is necessary, so that Cato can setup the initial knowledge (starting grid, your position, the initial strategy, etc.). If you join a session later than during the first lap, Cato will refuse to work. This behaviour can be changed with the setting "Strategist: Late Join" in the "Session Database", but the behaviour of Cato may be somewhat confusing.
Cato understands multi-class and multi-category races. Position evaluation and gap and lap time information will be always focused on your own class. Where it is necessary to mention, for example, the overall position, Cato will phrase it in a way, so that you understand, that information is related to the whole grid. Support tools like "Race Reports" also understand multi-class races and will give you related information with a class-specific focus as well, or you can choose, at which class you want to look, for example in a report. You can configure in the settings in the "Session Database", in what way Cato uses the information about the car classes and cup categories to partition the grid.
Cato shares many settings with Jona. Therefore consult the documentation on the race settings in the documentation for Jona for the general use of the settings tool and the "Session Database".
You will find settings for the race strategy analysis and simulation in the third tab of the settings tool.
A very special setting is the Autonomous Mode. If you set it to "Yes", it will allow the Assistants (especially the Strategist and the Engineer) to take their own decisions actions to guide you through the race without further confirmations needed from your side. Normally you will be asked to confirm each and every action taken by the Assistants, but this is not the case, when you choose "Yes". The default choice "Custom" let you even further customize the behaviour of the Assistants for every single decision and action. Take a look at the race settings in the "Session Database" and search there for "Confirm" to see all the available settings. Last, but not least, choosing No here will overwrite everything set in the "Session Database" and will force the Assistants to ask for confirmation each and every time. Further information can be found in the chapter about strategy handling.
You can also activate an automatic revision of a the currently chosen strategy with respect to the current race situation and current and maybe upcoming weather conditions. If a better strategy can be derived (depending on available data in the session database), Cato will inform you and you might decide to activate the new strategy (no confirmation is needed in autonomous mode, of course). Please note, that you can also ask the Strategist anytime to revise the strategy during a race, if you don't want to enable this automatism. Two different triggers can be defined for the recalculation, either automatic after a number of laps has been driven, or when an unplanned pitstop was neccessary. Please note, that the Strategist will never cancel the current strategy, if he cannot come up with a valid new one. Cancelling the current strategy must always be actively requested.
Using the fields at the end of the first group, you can customize the projection of current race positions into a model of future race standings. Cato will calculate this model each lap and will store the future standings for a given number of laps in the working memory and use them for strategy decisions. You can control the number of laps calculated for future standings by changing the value in the field Race positions. Greater numbers will yield better predictions, but take care, it might cost a lot of computing power. With the second field, Overtake Delta, you specify the number of seconds as time discount for each overtake for the passing and for the passed car, whereas you specify the percentage of track length in front of the car, which will be taken into account for traffic density analysis.
The second group of fields specify the time required for several pitstop activities, as well as the pitstop window, in which the best pitstop lap will be derived. With the value of Pitstop Delta, you supply the difference time needed for a normal pitstop (time for pit in and pit out but without any service time minus the time to pass the pit area on the track, i.e. Drive through vs. Drive by), The fields below specify the time required for the various pit services, like changing tyres, refueling, and so on, as well as these times are combined into an overall pitstop service time.
You can ask Cato to evaluate a couple of possible laps for an upcoming pitstop, either by using a voice command or the "RecommendPitstop" controller action. You can either ask to simulate a pitstop around a specific lap or you can simply ask for the best option for the next pitstop. In this case, the target lap is either taken from an active strategy or, when no such strategy has been created, it is determined depending on the amount of fuel left.
The Race Strategist will simulate the laps around the planned or requested pitstop lap (as long there is enough fuel to do it) and will try to optimize for undercut opportunities as well as the traffic density ahead after you re-enter the track. When you are satisfied with the prposed lap, Cato can handover the data to the Engineer, who will then prepare the pitstop.
Please note, that this simulation does not take possible pitstops or driving errors of your opponents into account. The simulation is based on the current positions and the average lap times of all drivers. If you want a more complex simulation, which is based on the Monte Carlo method, a team mate of you must run the "Team Center" and use the strategy tools, which are available on the Team Server. They can be used even when you are running a solo race.
Cato uses the position data gathered from the simulation game to form a complete overview of all drivers, their lap times and the development of their positions on track, even with different pit strategies. Using this knowledge, Cato can give you valuable information, but also can derive adapted race strategies, when your are stuck in traffic or are faced with other challenges.
The following statistical models are currently implemented:
-
Pre-Race strategy development
Cato can import a prepared strategy model at the start of a race session and will automatically call you to the pit and collaborate with Jona to give you a real life like race crew experience. Please see the dedicated chapter on the Strategy Workbench and also on Strategy Handling down below for more information.
You can configure in the race settings, that this predefined strategy will be revised actively by Cato depending on the current race situation or unplanned pitstops and so on.
-
Weather trend analysis
Cato will observe the weather development using the same rule set, that Jona uses. Thereby, Cato will notify you, when it is time to change the tyres, to achieve the best results. If you accept the recommendation, Cato will inform Jona to plan a pitstop with the best tyre compound for the upcoming conditions. Additionally, if Cato is currently following a predefined strategy, he will try to revise this strategy to take the new weather conditions into account.
-
Fuel availabilty and stint time calculation
Cato will observe the average fuel comsumption. Depending on the race situation and strategy requirements, Cato might suggest to save fuel to get one or two more laps from the available full. If you don't follow this recommendations, Cato will try to revise the strategy, to take this into account as well.
-
Standings and race position development
Using the position data gathered from the simulation game, Cato builds a knowledge of the pace of the various drivers. As a simple application of this knowledge, Cato can give you information about the current race positions and lap times of your opponents and the gaps between the cars. A more complex application will be a forecast of the race positions in a given time frame (see the next point).
-
Pitstop simulation
Also using the position data and a complex prediction model, Cato can determine the best lap for the next pitstop in a given pitstop window. The pitstop delta time as well as the service time is taken into account. The best pitstop lap will be selected based on position and the expected traffic after the pitstop.
Cato can use strategies that have been created with the "Strategy Workbench" during a race. When a race session starts, Cato looks for a currently defined strategy and checks, whether this strategy has been defined for the same car and track in the same conditions and whether the duration of the upcoming session is identical to that defined in the strategy. If all these aspects match, Cato will use the given strategy during the race. If there is a pitstop upcoming, Cato will actively inform you and will, as long as you accept it, hand over all settings for the pitstop to Jona for further handling. The pitstop will be planned and prepared for a specific lap and if you are in this lap, all settings will be entered into the Pitstop MFD of the current simulation without any necessary interaction from your side.
The Engineer, however, may revise parts of the request of the Strategist, especially for the last stint, because he has a much more precise understanding of the car state, fuel consumption, tyre wear, and so on. Those changes are only applied, if they are in accordance with the race rules.
Important: Cato keeps track of your pitstops, if you visit the pit after being called and with a preparation done by Jona. But Cato also checks whether the current lap fits to the strategy, if you visit the pit on your own. As long as the current lap is equal to the lap planned in the strategy +/- the number of laps defined as the pitstop window in the settings, it will count as a valid pitstop with regards to the strategy. Since most simulators do not flag a stop in the pits, whether it is a normal one, or whether it is for serving a penalty, this can lead to problems in those special cases, when you got a penalty and serve it right before the next planned pitstop. Cato tries to cope with that, but is not always successful. Sometimes, Cato will come up with a strategy revision after you have been in the pit for serving a penalty (depending on the conditions and the initial race rules), if you have enabled automatic strategy revisions in the settings. Be sure to check the revised strategy before accepting it, since Cato may count the penalty serving as a valid pitstop against the pitstop rules. If in doubt, reject the recommendation.
The currently active strategy might be dropped, if you do not adhere to the pitstop plan defined in the strategy, or it might be automatically revised - depending on the settings. To conform to the strategy, you must execute the pitstops in the laps as defined in the strategy +/- a few laps (default here is the pitstop variation window as defined in the race settings). If you think, that the predefined strategy is not of any use any more, because of a crash or changing weather conditions, you can cancel it actively with a voice command or by using "StrategyCancel" controller action.
Please note, that in a team race, the starting driver must be the one, which created and activated the strategy for the given event, or you musr use the "Instruct Strategist" command of the "Team Center" during the first laps to activate the strategy.
If you want to see all this in action in session in Assetto Corsa Competizione, take your time and watch this video:
The video has jump marks, so you can skip the boring time in between the interesting parts, where the Race Assistants take action.
If there is a need to revise the selected strategy later on, for example due to an accident with an unplanned repair pitstop or also due to severe wether changes, you can instruct the Strategist to recalculate and adjust the currently active strategy using a voice command or the "StrategyRecommend" controller action. In this case, the currently active strategy will be taken as a template incl. all original settings like pitstop rules, available tyre sets, and so on, but other aspects like starting fuel level or the current weather conditions, will be taken from the actual race situation. Based on this conditions a new strategy will be derived for the remaining race. The Strategist will inform you about the new strategy and will give you as much information as possible about the differences between the currently active and the new strategy. If it is not possible to calculate a new strategy, for example, when too few laps remain or if there is no telemetry data available for the requested weather conditions and tyre compound, the currently active strategy will be canceled upon your decision.
The weather forecast will be taken into account for the new strategy. If the currently mounted tyres are not suitable for the upcoming weather, the new strategy will start with a pitstop to change tyres. For additional information, how tyre compounds are chosen during weather changes, see the corresponding documentation about weather specific tyre compounds.
The Race Strategist will also keep track of all already used tyre sets and will consider this, when running a strategy simulation. However, this only covers tyre set usage in the current session. If there are tyre sets that already have been used in a previous session, for example in a qualifying run, this information is not available.
It is possible to enable a dynamic traffic simulation method, which uses a probabilistic method to predict future race development based on data collected during the race. You can enable or disable this method in the race settings or using default settings in "Session Database". Here you will also find a couple of settings to fine tune various aspects of the algorithm. These settings are similar to the options documented for the stratagy handling in the "Team Center, so please take a look there for more information about the Monte Carlo method.
Please note, that if you are running a race with Team Server support, a new strategy can be also created and activated using the strategy support of "Team Center", which gives you much more control over various aspects of the new strategy.
Good to know: The settings of the Optimizer will also be stored in the original strategy and the same settings will be used for all subsequent recalculations based on this particular strategy. This will be in general exactly what you want. But be aware, that the time needed (and therefore also the CPU consumption) to create a new strategy during a race will be the same as for the original strategy.
It is also possible to enable automatic revision of the current strategy in the race settings. You can specify that an automatic revision will occur after each couple of laps as well as after an unplanned pitstop. Cato will inform you whenever a better strategy has been identified and will give you as much information as possible about the differences between the currently active and the new strategy. You then can decide, that it should be activated for the rest of the session. Whether you can use this without sacrificing performance or introducing lags, depends on your PC configuration (see next advice). Please note, that with the current release, automatic revision will only bring up strategies which are to be considered better in terms of final number of laps or race time. This means that the strategist not automatically suggests a slower strategy, which may be necessary, for example, due to the onset of rain. The reason for this is the large number of edge cases in these situations. However, you can manually request a new strategy via voice command or controller action to override the current, probably better strategy.
Finally a word of a advice: The calculation of a strategy can be a time consuming and CPU intensive process, especially for long endurance races or if the Optimizer was used during the creation of the initially strategy, and, with even more impact, when the Monte Carlo traffic simulation method is used. This can result in framerate drops on weaker PC systems. So please check before an important race, whether you can use this functionality safely in your specific environment. But, even if your system is capable to handle the load, the recalculation may take some time, especially when high Optimizer settings were chosen during the creation of the original strategy.
Cato is also able to handle a Full Course Yellow condition, as long as a selected strategy is active. If you encounter a Full Course Yellow and if it is allowed to go to the pit during Full Course Yellow, you can ask the Strategist to evaluate whether a pitstop will be of benefit in this situation. There is a voice command and also a controller action available to trigger this evaluation. If the Strategist thinks that it is of benefit to perform an early pitstop, it will ask you and then directly calls you to the pit, while the Engineer is already preparing the pitstop.
Remarks:
- When Cato has created a new strategy with an immediate pitstop, only the number of pitstops are used to compare it against the currently active strategy, since in most cases, the benefits of performing a pitstop during Full Course Yellow outweighs a possible longer stint afterwards. Therefore, if the new strategy has the same number of remaining pitstops, it is considered valid and the Strategist will call you to the pit.
- Be careful, when using teh Full Course Yellow handling during the first few laps, since the calculation of average lap time and fuel consumption might not be exact then. Normally, the Asisstants will refuse to help you in the learning phase, but in case of Full Course Yellow this is overriden. As long as you have already collected data about the given car / track / weather condition, this information is used and you will be fine. Otherwise double-check the recommendation.
- If you have accepted the new strategy with an immediate pitstop, the Engineer will be informed without further notice. He will plan and prepare the pitstop immediately without waiting for the next crossing of the start/finish line, as usual. This is important, so that you can enter the pit at the end of the current lap.
If you have chosen Autonomous Mode in the Strategy tab of the "Race Settings" as described above, the Strategist and the Engineer will work together using the selected strategy in such a way, that no further interaction or decisions are needed from the drivers side regarding strategy updates, pitstop handling, and so on. Both Assistants will exhibit the same behaviour as in the non-autonomous mode, but they will not ask for confirmation, if they think that an action is required. Instead of asking for confirmation, the corresponding action is performed immediately and you are notified about that. As a result, the two Assistants now act as Agents, which feels very natural, very much like a real pit crew.
The Strateist will not ask for confirmation for the following actions:
- Give an introduction for the race strategy at the beginning of a session.
- Alter the strategy due to fuel requirements or changing conditions.
- Alter the strategy due to traffic requirements and possible undercut chances.
- Explain a changed strategy and the result of the corresponding traffic and position simulation.
- Calling you to the pit, thereby informing the Engineer to plan the pitstop.
Additionally, the Engineer will not ask for confirmation to prepare an already planned pitstop, when it is time to come to the pit.
Both Assistants will still ask for confirmation when actions are necessary due to weather changes, damages caused by accidents, and so on. It is possible to disable those confirmations as well, though. For this, take a look at the race settings in the "Session Database" and search there for "Confirm" to see all the available settings.
Cato uses the same AI kernel as Jona. In fact, large parts of the basic rule set is identical for both Assistants. Therefore, you can consult the technical information of Jona, if you want to dig deeper into the inner workings.
The complete position information is requested from the simulation gane every 10 seconds (this is the default, you can lower that in the settings, if your PC is fast enough). For each car, the data contains the current position, the last lap time and the fraction of the track which has been driven since the car last crossed the S/F line. The data acquisition is controlled by the "Race Strategist" plugin. For Assetto Corsa, RaceRoom Racing Experience, rFactor 2, Le Mans Ultimate, iRacing, Automobilista 2 and Project CARS 2 the data is requested from shared memory using the same data providers, which are used by Jona as well, whereas a special UDP client is used for Assetto Corsa Competizione (due to the asynchronous nature of the UDP interface for Assetto Corsa Competizione, the provided data is not always exact).
After the data has been gathered, it is then transfered to the Race Strategist process and loaded into the knowledge base, where the statistical models create several projections for future position development. Beside that, a historical copy is created for each lap.
[Position Data]
Driver.Car=1
Car.Count=20
Car.1.Car = Mazda MX-5 Cup
Car.1.Driver.Forname = The
Car.1.Driver.Nickname = TB
Car.1.Driver.Surname = BigO
Car.1.Laps = 2
Car.1.Lap.Running = 0
Car.1.Position = 1
Car.1.Time = 110650
Car.2.Car = Mazda MX-5 Cup
Car.2.Driver.Forname = Jimmy
Car.2.Driver.Nickname = JV
Car.2.Driver.Surname = Van Veen
Car.2.Laps = 1
Car.2.Lap.Running = 0.9514
Car.2.Position = 14
Car.2.Time = 123535
Car.3.Car = Mazda MX-5 Cup
Car.3.Driver.Forname = Jennifer
Car.3.Driver.Nickname = JY
Car.3.Driver.Surname = Young
Car.3.Laps = 1
Car.3.Lap.Running = 0.9827
Car.3.Position = 4
Car.3.Time = 117209
...
For some tips n tricks for the best voice recognition experience, see the corresponding chapter in the documentation of Jona.