Chapterبەش 3 · Lectureوانە 3.3

Logistics, Maintenance and Trainingلۆجستیک، چاککاری و ڕاهێنان

How AI helps an army, and an aid agency, get supplies to the right place at the right time: forecasting needs, predictive maintenance that fixes machines before they break, route optimisation and re-planning when a storm changes the roads, the World Food Programme's Optimus tool, and training with simulation, virtual reality, adaptive training and an AI opponent that gets harder each round. Simulation also tests the AI itself.

چۆن یارمەتیی سوپایەک، هەروەها دامەزراوەیەکی یارمەتیدان، دەدات بۆ ئەوەی پێداویستییەکان لە کاتی گونجاودا بگەنە شوێنی گونجاو: پێشبینیکردنی پێداویستییەکان، کە ئامێرەکان پێش تێکچوون چاک دەکاتەوە، و کاتێک زریانێک ڕێگاکان دەگۆڕێت، ئامرازی Optimusـی بەرنامەی خۆراکی جیهانی و ڕاهێنان بە و ، و ڕکابەرێکی کە لە هەر گەڕێکدا قورستر دەبێت. خودی تاقی دەکاتەوە.

15 min read18 خولەک خوێندنەوە · 14 stepsهەنگاو

What you will learnچی فێر دەبیت

  • Explain logistics and a supply chain, and how AI forecasts what each place will need.
  • Compare reactive, preventive and predictive maintenance, and describe the sensor → model → repair loop.
  • Treat route optimisation as a search on a weighted graph, and re-plan when a storm changes the costs.
  • Describe how simulation, virtual reality and adaptive training help people practise safely, and how an AI opponent gets harder.
  • Explain why simulation is also used to test AI systems before use, and why it is not enough on its own.
  • ڕوونی بکەیتەوە و چین و چۆن پێشبینیی پێداویستیی هەر شوێنێک دەکات.
  • دوای تێکچوون، و بەراورد بکەیت و خولی ← ← چاککردنەوە باس بکەیت.
  • وەک گەڕان لە ببینیت و کاتێک زریانێک کۆستەکان دەگۆڕێت پلان دابنێیتەوە.
  • باس بکەیت ، و چۆن یارمەتیی خەڵک دەدەن بە سەلامەتی ڕاهێنان بکەن و ڕکابەرێکی چۆن قورستر دەبێت.
  • ڕوونی بکەیتەوە بۆچی بۆ تاقیکردنەوەی سیستەمەکانی پێش بەکارهێنان بەکار دێت و بۆچی بەتەنیا بەس نییە.

The right supplies, in the right place, at the right timeپێداویستیی گونجاو، لە شوێنی گونجاو، لە کاتی گونجاودا

An army cannot move without logistics: getting fuel, food, water, medicine, spare parts and equipment, from radios to vehicles and weapons, to the right place at the right time. The goods travel along a supply chain: from makers and suppliers, through a main depot and hubs, to the camps, field hospitals and workshops that use them.

سوپا بەبێ ناجووڵێت: گەیاندنی سووتەمەنی، خواردن، ئاو، دەرمان، پارچەی یەدەگ و کەلوپەل، لە بێتەلەوە تا ئۆتۆمبێل و چەک، بۆ شوێنی گونجاو لە کاتی گونجاودا. کاڵاکان بە دەڕۆن: لە دروستکەر و دابینکەرانەوە، بە کۆگایەکی سەرەکی و ناوەندەکانی دابەشکردندا، بۆ ئۆردوگاکان، نەخۆشخانە مەیدانییەکان و وەرشەکان کە بەکاریان دەهێنن.

AI helps in three places: it forecasts what each place will need, predicts which machines will break, and plans routes for the trucks. People still decide what is sent, where and when. UNIDIR, the UN institute for disarmament research, counts logistics, including predictive maintenance, as one of the four main groups of military AI uses, beside command and control, information and training.

لە سێ شوێندا یارمەتی دەدات: پێشبینیی ئەوە دەکات کە هەر شوێنێک پێویستی بە چییە، پێشبینیی ئەوە دەکات کام ئامێر تێک دەچێت و ڕێگا بۆ بارهەڵگرەکان پلان دەکات. هێشتا مرۆڤ بڕیار دەدات چی، بۆ کوێ و کەی بنێردرێت. UNIDIR، دامەزراوەی نەتەوە یەکگرتووەکان بۆ توێژینەوەی چەکداماڵین، ، لەوانەش ، بە یەکێک لە چوار گرووپە سەرەکییەکەی بەکارهێنانی دادەنێت، لەتەک فەرماندەیی و کۆنترۆڵ، زانیاری و ڕاهێنان.

A supply chain from left to right. Makers and suppliers (a factory) send goods to a main depot (a warehouse). From the depot, trucks carry them to two hubs, north and south, and from the hubs to a camp (a tent), a field hospital (a tent with a cross) and a repair workshop (a wrench). Purple tags show where AI helps: forecast needs (at the hubs), plan routes (on the roads) and predict breakdowns (at the workshop). A strip lists what moves along the chain: fuel, food, water, medicine, spare parts, equipment.زنجیرەیەکی دابینکردن لە چەپەوە بۆ ڕاست. دروستکەر و دابینکەران (کارگەیەک) کاڵا بۆ کۆگایەکی سەرەکی دەنێرن. لە کۆگاکەوە بارهەڵگرەکان دەیبەن بۆ دوو ناوەندی دابەشکردن (باکوور و باشوور) و لە ناوەندەکانەوە بۆ ئۆردوگایەک (چادرێک)، نەخۆشخانەیەکی مەیدانی (چادرێک بە خاچێکەوە) و وەرشەیەکی چاککردنەوە (ئاچارێک). نیشانە مۆرەکان پیشان دەدەن ژیریی دەستکرد لە کوێ یارمەتی دەدات: پێشبینیی پێداویستی (لە ناوەندەکان)، پلانی ڕێگا (لەسەر ڕێگاکان) و پێشبینیی تێکچوون (لە وەرشەکە). ڕیزێک ئەوەی بە زنجیرەکەدا دەڕوات دەژمێرێت: سووتەمەنی، خواردن، ئاو، دەرمان، پارچەی یەدەگ، کەلوپەل.
Logistics is the whole chain, not only the trucks. هەموو زنجیرەکەیە، نەک تەنیا بارهەڵگرەکان.CêIT

The right supplies, in the right place, at the right timeپێداویستیی گونجاو، لە شوێنی گونجاو، لە کاتی گونجاودا

Forecasting what each place will needپێشبینیکردنی ئەوەی هەر شوێنێک پێویستی پێیەتی

To send the right amount, planners must look into the future: how much fuel, food and water will each place use next week? A forecasting model learns from past use and adds what is known now: how many people are there, the weather forecast, distances and planned activities. This is called predictive analytics, here used for supplies.

بۆ ناردنی بڕی گونجاو، پلاندانەران دەبێت سەیری داهاتوو بکەن: هەر شوێنێک هەفتەی داهاتوو چەند سووتەمەنی، خواردن و ئاو بەکار دەهێنێت؟ پێشبینی لە بەکارهێنانی پێشووەوە فێر دەبێت و ئەوەی ئێستا دەزانرێت زیادی دەکات: ژمارەی خەڵک لەوێ، پێشبینیی کەشوهەوا، دووری و چالاکییە پلانکراوەکان. بەم کارە دەوترێت شیکاریی پێشبینیکەر، کە لێرەدا بۆ پێداویستییەکان بەکار دێت.

Left: four inputs, past daily use, number of people, weather forecast and planned activities, feed a forecasting model. Right: a bar chart of the drinking water one camp needs for the next 7 days, in thousand litres per day, each bar with a line showing a range. Days 1 to 3 are about 22 to 23; days 4 and 5 are hot days at about 28 to 29, coloured orange; days 6 and 7 fall to about 23 to 24. The range lines grow longer for later days. Caption: the further ahead we look, the less sure the forecast is. The numbers are an example.لای چەپ: چوار نێوخستە، بەکارهێنانی ڕۆژانەی پێشوو، ژمارەی خەڵک، پێشبینیی کەشوهەوا و چالاکییە پلانکراوەکان، دەچنە ناو مۆدێلی پێشبینی. لای ڕاست: هێڵکارییەکی ستوونی بۆ ئەو ئاوی خواردنەوەیەی ئۆردوگایەک بۆ حەوت ڕۆژی داهاتوو پێویستی دەبێت، بە هەزار لیتر لە ڕۆژێکدا؛ هەر ستوونێک هێڵێکی هەیە کە کەمترین و زۆرترین پیشان دەدات. ڕۆژەکانی 1 تا 3 نزیکەی 22 تا 23ن؛ ڕۆژەکانی 4 و 5 ڕۆژی گەرمن بە نزیکەی 28 تا 29، بە ڕەنگی پرتەقاڵی؛ ڕۆژەکانی 6 و 7 دادەبەزن بۆ نزیکەی 23 تا 24. ئەو هێڵانە بۆ ڕۆژە دوورترەکان درێژتر دەبن. ڕوونکردنەوە: هەتا دوورتر بڕوانین، پێشبینییەکە کەمتر دڵنیایە. ژمارەکان نموونەن.
A good forecast gives a range, not one magic number.پێشبینییەکی باش کەمترین و زۆرترین دەدات، نەک یەک ژمارەی جادوویی.CêIT

A good forecast gives a range and is checked against what really happens. If a flood closes a road or more people arrive, the forecast is updated. Too little water is dangerous; too much wastes trucks and fuel.

پێشبینییەکی باش کەمترین و زۆرترین بڕی چاوەڕوانکراو دەدات و بەراورد دەکرێت بەوەی بەڕاستی ڕوو دەدات. ئەگەر لافاوێک ڕێگایەک دابخات یان خەڵکی زیاتر بگەن، پێشبینییەکە نوێ دەکرێتەوە. ئاوی کەم مەترسیدارە؛ ئاوی زیاد بارهەڵگر و سووتەمەنی بەفیڕۆ دەدات.

Fixing a machine before it breaksچاککردنەوەی ئامێرێک پێش ئەوەی تێک بچێت

Trucks, generators, aircraft and water pumps tend to break at the worst moment. Predictive maintenance tries to see the failure coming. Sensors measure things like vibration, temperature, oil quality and running hours. A model learns what these readings look like in the weeks before a part fails, and it warns: 'this pump will probably fail within about 5 days'. Mechanics then repair it at a planned time, before the breakdown.

بارهەڵگر، مۆلیدە، فڕۆکە و پەمپی ئاو زۆرجار لە خراپترین کاتدا تێک دەچن. هەوڵ دەدات تێکچوونەکە پێش ئەوەی ڕوو بدات ببینێت. شتی وەک لەرینەوە، پلەی گەرمی، باشیی ڕۆن و کاتژمێرەکانی کارکردن دەپێون. فێر دەبێت ئەم خوێندنەوانە لە هەفتەکانی پێش تێکچوونی پارچەیەکدا چۆنن و ئاگادار دەکاتەوە: «ئەم پەمپە بە ئەگەری زۆر لە ماوەی نزیکەی پێنج ڕۆژدا تێک دەچێت». پاشان میکانیکییەکان لە کاتێکی پلانکراودا، پێش تێکچوونەکە، چاکی دەکەنەوە.

Animation in five steps on a chart of a water pump's vibration over 30 days. A shaded band at the top is the breakdown zone, and a dashed line below it is the warning line where a repair should be planned. Days 1 to 14: the line is flat at about 3 mm/s. Then a worn bearing makes it rise. On day 22 it crosses the warning line, and the model's dashed forecast reaches the breakdown zone about 5 days later. On day 24 a wrench appears: the mechanics change the bearing, and the line drops back to about 2.6 mm/s. Last step: no breakdown and no surprise; the pump keeps working.
Animation in five steps on a chart of a water pump's vibration over 30 days. A shaded band at the top is the breakdown zone, and a dashed line below it is the warning line where a repair should be planned. Days 1 to 14: the line is flat at about 3 mm/s. Then a worn bearing makes it rise. On day 22 it crosses the warning line, and the model's dashed forecast reaches the breakdown zone about 5 days later. On day 24 a wrench appears: the mechanics change the bearing, and the line drops back to about 2.6 mm/s. Last step: no breakdown and no surprise; the pump keeps working.
The data says 'soon'; people plan the repair before the breakdown. دەڵێن «بەم زووانە»؛ مرۆڤ پێش تێکچوون پلانی چاککردنەوە دادەنێت.CêIT

A real example: in 2023 the US Air Force made an AI tool called PANDA its official system for this kind of maintenance. It combines onboard sensor data, flight data and maintenance and supply records, and maintenance and engineering analysts review its recommendations before acting.

نموونەیەکی ڕاستەقینە: لە ⁦2023⁩دا هێزی ئاسمانیی ئەمریکا ئامرازێکی بە ناوی PANDA کردە سیستەمی فەرمیی خۆی بۆ ئەم جۆرە . ئەم ئامرازە سەر فڕۆکە، فڕین و تۆمارەکانی و پێداویستی پێکەوە کۆ دەکاتەوە و شیکەرەوەکانی و ئەندازیاری پێش ئەوەی کار بکەن پێداچوونەوە بە پێشنیارەکانیدا دەکەن.

Fixing a machine before it breaksچاککردنەوەی ئامێرێک پێش ئەوەی تێک بچێت

Three ways to look after a machineسێ ڕێگا بۆ ئاگالێبوونی ئامێرێک

The US Department of Energy describes the basic approaches in a few words: reactive maintenance means fix or replace it when it breaks; preventive maintenance means time-based care on a fixed timetable; predictive maintenance means fix it before it breaks, based on the machine's real condition.

وەزارەتی وزەی ئەمریکا ڕێگا سەرەکییەکان بە چەند وشەیەک باس دەکات: دوای تێکچوون واتە کاتێک تێکچوو چاکی بکەرەوە یان بیگۆڕە؛ واتە ئاگالێبوون بەپێی کات و بە خشتەیەکی جێگیر؛ واتە پێش تێکچوون چاکی بکەرەوە، بەپێی دۆخی ڕاستەقینەی ئامێرەکە.

Three rows over 1,000 running hours. Reactive: the machine runs until about 700 hours, breaks down (a red cross) and stands still for days before it is repaired. Preventive: a wrench every 250 hours, at 250, 500 and 750; some parts are replaced while still good. Predictive: a sensor line stays flat, starts rising after about 480 hours, crosses a dashed warning line near 630 hours, and the machine is repaired at about 700 hours, when the sensor data says so, with no breakdown.سێ ڕیز لە ماوەی 1,000 کاتژمێر کارکردندا. چاککاریی دوای تێکچوون: ئامێرەکە تا نزیکەی 700 کاتژمێر کار دەکات، تێک دەچێت (خاچێکی سوور) و چەند ڕۆژێک دەوەستێت پێش ئەوەی چاک بکرێتەوە. چاککاریی پێشوەختە: هەر 250 کاتژمێرێک ئاچارێک، لە 250، 500 و 750دا؛ هەندێک پارچە دەگۆڕدرێن کە هێشتا باش بوون. چاککاریی پێشبینیکەر: هێڵی هەستەوەرێک ڕێک دەمێنێتەوە، دوای نزیکەی 480 کاتژمێر بەرز دەبێتەوە، نزیکەی 630 کاتژمێر هێڵێکی پچڕپچڕی ئاگادارکردنەوە دەبڕێت و ئامێرەکە نزیکەی 700 کاتژمێر چاک دەکرێتەوە، کاتێک دراوەی هەستەوەرەکان دەڵێن، بەبێ تێکچوون.
Same machine, three strategies: wait, follow the calendar, or follow the data.هەمان ئامێر، سێ ستراتیژ: چاوەڕوانی، پەیڕەوکردنی ڕۆژژمێر یان پەیڕەوکردنی .CêIT
ApproachڕێگاGoodباشیWeakلاوازی
Reactive دوای تێکچوونno planning or sensors neededپێویستی بە پلان یان نییەsurprise breakdowns, long stops, dangerous in the fieldتێکچوونی چاوەڕواننەکراو، وەستانی درێژ، مەترسیدار لە مەیداندا
Preventivefewer surprises; easy to planسەرسوڕمانی کەمتر؛ پلاندانانی ئاسانgood parts are replaced too early; some failures still happen between visitsپارچەی باش زوو دەگۆڕدرێت؛ هێشتا هەندێک تێکچوون لە نێوان سەردانەکاندا ڕوو دەدات
Predictiverepairs only when needed; fewer breakdownsچاککردنەوە تەنیا کاتێک پێویستە؛ تێکچوونی کەمترneeds sensors, clean data and a good model; false alarms are possibleپێویستی بە ، پاک و باش هەیە؛ ئاگادارکردنەوەی هەڵە دەشێت ڕوو بدات

Finding the best routeدۆزینەوەی باشترین ڕێگا

Route optimisation means finding the best way through a road network. The map is drawn as a weighted graph: places are nodes, roads are edges, and each road has a cost. The cost can be minutes, litres of fuel or a risk score, or a mix of them. A search algorithm such as A* search then finds the cheapest path.

واتە دۆزینەوەی باشترین ڕێگا بە ناو تۆڕێکی ڕێگاوباندا. نەخشەکە وەک گرافی کێشدار دەکێشرێت: شوێنەکان دەبنە ، ڕێگاکان دەبنە و هەر ڕێگایەک کۆستی خۆی هەیە. کۆستەکە دەکرێت خولەک بێت، لیتری سووتەمەنی یان نمرەی ، یان تێکەڵەیەک لەوانە. پاشان گەڕانی وەک گەڕانی ⁦A*⁩ هەرزانترین دەدۆزێتەوە.

A road map drawn as a weighted graph. Nodes: Depot, A, B, C, D and Camp. Roads with their minutes: Depot–A 15, Depot–B 20, A–B 10, A–C 20, B–D 25, C–D 10, C–Camp 20, D–Camp 15. The cheapest route is highlighted: Depot → A → C → Camp = 15 + 20 + 20 = 55 minutes.نەخشەی ڕێگاوبان وەک گرافی کێشدار. نۆدەکان: کۆگا، A، B، C، D و ئۆردوگا. ڕێگاکان و خولەکەکانیان: کۆگا بۆ A 15، کۆگا بۆ B 20، A بۆ B 10، A بۆ C 20، B بۆ D 25، C بۆ D 10، C بۆ ئۆردوگا 20، D بۆ ئۆردوگا 15. هەرزانترین ڕێگا نیشانە کراوە: لە کۆگاوە بە A و Cـدا بۆ ئۆردوگا، 15 + 20 + 20 = 55 خولەک.
Each road's cost is in minutes here; the cheapest route takes 55 minutes.لێرەدا کۆستی هەر ڕێگایەک بە خولەکە؛ هەرزانترین ڕێگا ⁦55⁩ خولەک دەخایەنێت.CêIT

'Best' depends on the cost you choose. Count only minutes and you get the fastest route; add a penalty for dangerous roads and you get the safest one that is still quick. Choosing the cost is a human decision: it says what matters most.

«باشترین» بەندە بەو کۆستەی هەڵیدەبژێریت. ئەگەر تەنیا خولەک بژمێریت، خێراترین ڕێگا وەردەگریت؛ ئەگەر سزایەک بۆ ڕێگا مەترسیدارەکان زیاد بکەیت، سەلامەتترین ڕێگا وەردەگریت کە هێشتا خێرایە. هەڵبژاردنی کۆست بڕیارێکی مرۆییە: دەڵێت چی لە هەمووی گرنگترە.

Finding the best routeدۆزینەوەی باشترین ڕێگا

A storm changes the costsزریانێک کۆستەکان دەگۆڕێت

Now a storm arrives while the truck is on the road. One road floods and another becomes slow. In the graph their costs go up, and the flooded road is removed. The old 'best' route is no longer best, so the planner re-plans: it searches again from where the truck is now, with the new costs.

ئێستا لە کاتێکدا بارهەڵگرەکە لە ڕێگادایە زریانێک دێت. ڕێگایەک لافاو دەیگرێت و ڕێگایەکی تر هێواش دەبێت. لە کۆستەکانیان بەرز دەبنەوە و ڕێگا لافاوگرتووەکە لادەبرێت. «باشترین» ڕێگای پێشوو ئیتر باشترین نییە، بۆیە پلاندانەر دەکات: دووپاتە دەگەڕێت، لەو شوێنەوەی بارهەڵگرەکە ئێستا لێیەتی، بە کۆستە نوێیەکان.

Animation in seven steps on the same road map. 1: the plan before leaving, Depot → A → C → Camp, 55 minutes. 2: after 15 minutes the truck reaches A. 3: a storm cloud appears over C; road C–Camp is flooded and crossed out, and road A–C now takes 45 minutes instead of 20. 4: re-planning from A finds A → B → D → Camp, 10 + 25 + 15 = 50 minutes. 5 and 6: the truck follows the new route through B and D. 7: it arrives after 65 minutes, 10 minutes later than planned, but safe.
Animation in seven steps on the same road map. 1: the plan before leaving, Depot → A → C → Camp, 55 minutes. 2: after 15 minutes the truck reaches A. 3: a storm cloud appears over C; road C–Camp is flooded and crossed out, and road A–C now takes 45 minutes instead of 20. 4: re-planning from A finds A → B → D → Camp, 10 + 25 + 15 = 50 minutes. 5 and 6: the truck follows the new route through B and D. 7: it arrives after 65 minutes, 10 minutes later than planned, but safe.
Re-planning starts from the truck's current position, with the new costs. لە شوێنی ئێستای بارهەڵگرەکەوە دەست پێ دەکات، بە کۆستە نوێیەکان.CêIT

The hoped-for result is faster deliveries with fewer breakdowns. That is an expected benefit, not a promise: it depends on fresh road and weather reports, and on drivers who can report what they see. When the data is wrong, the plan is wrong.

ئەنجامی چاوەڕوانکراو گەیاندنی خێراتر و تێکچوونی کەمترە. ئەمە سوودێکی چاوەڕوانکراوە، نەک بەڵێن: بەندە بە ڕاپۆرتی نوێی ڕێگا و کەشوهەوا و بە شۆفێرانێک کە بتوانن ئەوەی دەیبینن ڕاپۆرت بکەن. کاتێک هەڵە بێت، پلانەکەش هەڵەیە.

Finding the best routeدۆزینەوەی باشترین ڕێگا

The same tools feed peopleهەمان ئامرازەکان خەڵک تێر دەکەن

The same mathematics serves aid agencies. The UN World Food Programme (WFP) built Optimus, a decision-support tool. It uses data and mathematical models to suggest which foods to put in a food basket so that it is nutritious and affordable, where to buy them and how to deliver them, within limits such as funding and delivery times. So far WFP has used it in 44 of its operations (as of November 2026); in 2023 it reached 1.7 million people, and in Haiti it cut operating costs by US$273,000.

هەمان بیرکاری خزمەتی دامەزراوەکانی یارمەتیدان دەکات. بەرنامەی خۆراکی جیهانیی نەتەوە یەکگرتووەکان ⁦(WFP)⁩ ئامرازێکی بە ناوی Optimus دروست کرد کە . و مۆدێلی بیرکاری بەکار دەهێنێت بۆ پێشنیارکردنی ئەوەی کام خواردن بخرێتە سەبەتەی خۆراکەوە تا هەم بە سوود بێت و هەم هەرزان، لە کوێ بکڕدرێن و چۆن بگەیەنرێن، لە چوارچێوەی سنووری وەک پارە و کاتی گەیاندن. تا تشرینی دووەمی ⁦2026⁩، WFP لە ⁦44⁩ کاری خۆیدا بەکاری هێناوە؛ لە ⁦2023⁩دا گەیشتە ⁦1.7⁩ ملیۆن کەس و لە هایتی ⁦273,000⁩ دۆلاری ئەمریکی لە تێچووی کارەکان کەم کردەوە.

Four inputs on the left, nutrition needs, food prices in each market, routes and delivery times, and the budget, flow into a box: Optimus, where data and mathematical models suggest plans. Arrows lead to three outputs: food basket options, where to buy the food, and a delivery plan. Below, WFP planners compare the options and decide. A strip at the bottom: used in 44 WFP operations; reached 1.7 million people in 2023.چوار نێوخستە لە لای چەپ، پێداویستیی خۆراکی، نرخی خواردن لە هەر بازاڕێک، ڕێگاکان و کاتی گەیاندن و بودجە، دەچنە ناو چوارگۆشەیەک: Optimus، کە تێیدا دراوە و مۆدێلی بیرکاری پلان پێشنیار دەکەن. تیرەکان دەچن بۆ سێ ئەنجام: هەڵبژاردەکانی سەبەتەی خۆراک، لە کوێ خواردنەکە بکڕدرێت و پلانێکی گەیاندن. لە خوارەوە پلاندانەرانی WFP هەڵبژاردەکان بەراورد دەکەن و بڕیار دەدەن. ڕیزێک لە خوارەوە: لە 44 کاری WFPـدا بەکارهاتووە؛ لە 2023دا گەیشتە 1.7 ملیۆن کەس.
Optimus suggests; WFP's people decide.Optimus پێشنیار دەکات؛ خەڵکی WFP بڕیار دەدەن.CêIT

Optimus suggests and WFP's teams decide. That is the pattern of every good decision support system, in war and in peace; the lesson on human control looks at it closely.

Optimus پێشنیار دەکات و تیمەکانی WFP بڕیار دەدەن. ئەمە شێوازی هەر سیستەمێکی پشتگیریی بڕیاردانی باشە، لە جەنگ و لە ئاشتیدا؛ وانەی کۆنترۆڵی مرۆڤ بە وردی سەیری دەکات.

Practising safely: simulation and virtual realityڕاهێنانی سەلامەت: هاوشێوەسازی و ڕاستیی گریمانەیی

Real exercises are expensive, and some situations are too dangerous or too rare to practise for real: an engine fire in flight, a convoy stuck in a flood, a collapsed building. A simulation recreates them on a computer. Trainees can sit in a simulator, a copy of a cockpit or a driver's seat, or wear a virtual reality headset, and practise as often as they need without danger. The three kinds of immersive technology are compared in VR, AR or mixed reality?.

ڕاهێنانی ڕاستەقینە گرانە و هەندێک بارودۆخ زۆر مەترسیدار یان زۆر دەگمەنن بۆ ئەوەی بە ڕاستی ڕاهێنانیان لەسەر بکرێت: ئاگرگرتنی بزوێنەر لە کاتی فڕیندا، کاروانێک کە لە لافاودا گیر بووە، باڵەخانەیەکی ڕووخاو. ئەمانە لەسەر دروست دەکاتەوە. ڕاهێنراوەکان دەتوانن لە دابنیشن، کە کۆپییەکی کابینەی فڕۆکە یان کورسیی شۆفێرە، یان لە سەر بکەن و هەر چەند جار پێویست بێت بەبێ مەترسی ڕاهێنان بکەن. ئەم سێ جۆرە تەکنەلۆجیایە لە کۆرسی بەراورد کراون (VR، AR یان ڕاستیی تێکەڵ).

AI makes simulations more useful in two ways: it adapts the training to each person, and it plays the other side, so trainees have someone to practise against. The next two screens show each.

بە دوو ڕێگا بەسوودتر دەکات: ڕاهێنانەکە لەگەڵ هەر کەسێک دەگونجێنێت و لایەنی بەرامبەر دەگێڕێت، بۆ ئەوەی ڕاهێنراوەکان کەسێکیان هەبێت بەرامبەری ڕاهێنان بکەن. دوو بەشی داهاتوو هەر یەکەیان پیشان دەدەن.

Practising safely: simulation and virtual realityڕاهێنانی سەلامەت: هاوشێوەسازی و ڕاستیی گریمانەیی

Training that adapts to each personڕاهێنانێک کە خۆی لەگەڵ هەر کەسێک دەگونجێنێت

In a normal class, everyone gets the same lesson. Adaptive training watches each learner's results and changes the next task: harder after a success, easier and with a hint after a mistake, and more practice on whatever skill is weak.

لە پۆلێکی ئاساییدا هەموو کەس هەمان وانە وەردەگرێت. سەیری ئەنجامەکانی هەر فێرخوازێک دەکات و ئەرکی داهاتوو دەگۆڕێت: دوای سەرکەوتن قورستر، دوای هەڵە ئاسانتر و لەگەڵ ڕێنمایی، هەروەها ڕاهێنانی زیاتر لەسەر هەر کارامەییەک کە لاوازە.

A chart of difficulty level (1 to 5) over six training rounds. Round 1: level 1, success. Round 2: level 2, success. Round 3: level 3, a mistake (orange cross). Round 4: back to level 2 with a hint, success (purple). Round 5: level 3, success. Round 6: level 4, success. The legend says: success means a harder task next; a mistake means an easier task next; with a hint, then up again when it works.هێڵکارییەکی ئاستی قورسی (1 تا 5) لە ماوەی شەش گەڕی ڕاهێناندا. گەڕی 1: ئاستی 1، سەرکەوتن. گەڕی 2: ئاستی 2، سەرکەوتن. گەڕی 3: ئاستی 3، هەڵە (خاچێکی پرتەقاڵی). گەڕی 4: گەڕانەوە بۆ ئاستی 2 لەگەڵ ڕێنمایی، سەرکەوتن (مۆر). گەڕی 5: ئاستی 3، سەرکەوتن. گەڕی 6: ئاستی 4، سەرکەوتن. ڕوونکەرەوەکە دەڵێت: سەرکەوتن واتە دواتر ئەرکێکی قورستر؛ هەڵە واتە دواتر ئەرکێکی ئاسانتر؛ لەگەڵ ڕێنمایی، پاشان کە سەرکەوت بەرز دەبێتەوە.
The path is different for every learner, and that is the point.ڕێگاکە بۆ هەر فێرخوازێک جیاوازە، مەبەستیش هەر ئەوەیە.CêIT
Like a good driving teacherوەک مامۆستایەکی باشی شۆفێری

A good driving teacher notices that you struggle with parking and makes you park again, but lets you skip what you already do well.

مامۆستایەکی باشی شۆفێری تێدەگات کە لە پارککردندا ئاستەنگت هەیە و دووپاتە وات لێدەکات پارک بکەیت، بەڵام ڕێگەت پێ دەدات ئەوەی بە باشی دەیکەیت تێپەڕێنیت.

Where the analogy stops workingسنووری ئەم نموونەیە

A teacher also notices tiredness, fear and mood; software sees only the results it measures.

مامۆستا هەروەها ماندوویی، ترس و حاڵی دەروونیت دەبینێت؛ تەنیا ئەو ئەنجامانە دەبینێت کە دەیانپێوێت.

The US Army's research centre publishes GIFT, a framework for building adaptive tutors: one part estimates what the learner knows, and another decides when to give feedback or a hint.

ناوەندی توێژینەوەی سوپای ئەمریکا GIFT بڵاو دەکاتەوە، کە چوارچێوەیەکە بۆ دروستکردنی مامۆستای کە خۆی دەگونجێنێت: بەشێکی هەڵدەسەنگێنێت فێرخوازەکە چی دەزانێت و بەشێکی تر بڕیار دەدات کەی ڕاوبۆچوون یان ڕێنمایی بدات.

Practising safely: simulation and virtual realityڕاهێنانی سەلامەت: هاوشێوەسازی و ڕاستیی گریمانەیی

An AI opponent that gets harderڕکابەرێکی ژیریی دەستکرد کە قورستر دەبێت

In many simulations the computer plays the other side, the opposing team of an exercise. Each round it can use smarter tactics, so the trainee always has a challenge. You met the same idea in the AI course: in Minimax difficulty levels, searching more moves ahead makes a harder opponent. Modern game AI also learns by trial and error (reinforcement learning).

لە زۆر لایەنی بەرامبەر دەگێڕێت، واتە تیمی ڕکابەری ڕاهێنانەکە. لە هەر گەڕێکدا دەتوانێت تاکتیکی زیرەکتر بەکار بهێنێت، بۆ ئەوەی ڕاهێنراوەکە هەمیشە ئاستەنگێکی لە بەردەمدا بێت. هەمان بیرۆکەت لە کۆرسی بینیوە: لە ئاستەکانی قورسیی Minimax، گەڕان بە جووڵەی زیاتر بەرەو پێشەوە ڕکابەرێکی قورستر دروست دەکات. یارییە هاوچەرخەکانیش بە تاقیکردنەوە و هەڵە فێر دەبێت (فێربوونی بەهێزکردنەوە).

Animation in five steps. Left, a trainee wearing a VR headset, with a skill bar; right, an AI opponent with a level badge; in the middle, a 5 × 5 board for the simulated exercise with a blue and an orange token. Round 1: level 1, always the same moves; the trainee learns the basics (skill 35%). Round 2: level 2, it looks one move ahead (skill 50%). Round 3: level 3, it looks several moves ahead (skill 65%). Round 4: level 4, it changes its tactics, so the trainee must adapt too (skill 80%). Last caption: harder each round, better each round, and mistakes cost nothing in a simulation.
Animation in five steps. Left, a trainee wearing a VR headset, with a skill bar; right, an AI opponent with a level badge; in the middle, a 5 × 5 board for the simulated exercise with a blue and an orange token. Round 1: level 1, always the same moves; the trainee learns the basics (skill 35%). Round 2: level 2, it looks one move ahead (skill 50%). Round 3: level 3, it looks several moves ahead (skill 65%). Round 4: level 4, it changes its tactics, so the trainee must adapt too (skill 80%). Last caption: harder each round, better each round, and mistakes cost nothing in a simulation.
The opponent grows with the trainee.ڕکابەرەکە لەگەڵ ڕاهێنراوەکەدا گەشە دەکات.CêIT

The trainee may wear a VR headset and practise decisions again and again. The goal is not to 'win a game' but to build safe habits before real life demands them.

ڕاهێنراوەکە دەشێت لە سەر بکات و بڕیاردان چەندین جار دووپاتە بکاتەوە. ئامانج «بردنەوەی یارییەک» نییە، بەڵکو دروستکردنی ڕاهاتنی سەلامەتە پێش ئەوەی ژیانی ڕاستەقینە داوایان بکات.

Simulation tests the AI tooهاوشێوەسازی ژیریی دەستکردیش تاقی دەکاتەوە

Simulation trains people, and it also tests machines. Before an AI system is used, it can face thousands of simulated situations: rain, night, a broken sensor, a closed road, people nearby. Engineers measure its mistakes, fix them and test again, and only then try it in the real world with people watching.

خەڵک ڕادەهێنێت و ئامێریش تاقی دەکاتەوە. پێش ئەوەی سیستەمێکی بەکار بهێنرێت، دەکرێت ڕووبەڕووی هەزاران بارودۆخی هاوشێوەسازیکراو ببێتەوە: باران، شەو، تێکچوو، ڕێگایەکی داخراو، خەڵک لە نزیکەوە. ئەندازیاران هەڵەکانی دەپێون، چاکیان دەکەنەوە و دووپاتە تاقیی دەکەنەوە و تەنیا دوای ئەوە، بە سەرپەرشتیی مرۆڤ، لە جیهانی ڕاستەقینەدا تاقیی دەکەنەوە.

On 4 September 2026 the government experts who discuss autonomous weapons under the Convention on Certain Conventional Weapons agreed by consensus that such weapons should be tested and evaluated 'including adequately realistic simulations', to see how they will perform where they will be used. Their report goes to the convention's Review Conference on 16–20 November 2026.

لە 4ی ئەیلوولی ⁦2026⁩دا شارەزایانی حکوومەتەکان، کە لە چوارچێوەی باسی دەکەن، بە کۆدەنگی ڕێککەوتن کە دەبێت ئەم جۆرە چەکانە تاقی بکرێنەوە و هەڵبسەنگێنرێن، «لەوانەش بە بەس نزیک لە ڕاستی»، بۆ ئەوەی ببینرێت لەو شوێنەی بەکار دەهێنرێن چۆن کار دەکەن. ڕاپۆرتەکەیان دەچێتە بەردەم کۆنگرەی پێداچوونەوەی پەیماننامەکە لە ⁦16⁩ تا ⁦20⁩ی تشرینی دووەمی ⁦2026⁩.

A testing loop. An AI system goes into a simulated world with thousands of situations, for example rain, night, a broken sensor, a closed road and people nearby. Next, the mistakes are measured (false alarms and misses); then the system is fixed and trained again, and it goes back into the simulated world. When it is good enough, an arrow leads out to a real trial with people watching, and then to use with continued monitoring. A strip at the bottom quotes the CCW experts (September 2026): test and evaluate, 'including adequately realistic simulations'.خولێکی تاقیکردنەوە. سیستەمێکی ژیریی دەستکرد دەچێتە ناو جیهانێکی هاوشێوەسازیکراو بە هەزاران بارودۆخەوە، بۆ نموونە باران، شەو، هەستەوەرێکی تێکچوو، ڕێگای داخراو و خەڵک لە نزیکەوە. پاشان هەڵەکان دەپێورێن (ئاگادارکردنەوەی هەڵە و لەدەستچوون)، دواتر سیستەمەکە چاک دەکرێتەوە و دووپاتە ڕادەهێنرێتەوە و دەگەڕێتەوە بۆ جیهانە هاوشێوەسازیکراوەکە. کاتێک بەس باش بوو، تیرێک دەیباتە دەرەوە بۆ تاقیکردنەوەیەکی ڕاستەقینە بە سەرپەرشتیی مرۆڤ و پاشان بۆ بەکارهێنان لەگەڵ پشکنینی بەردەوام. ڕیزێک لە خوارەوە قسەی شارەزایانی CCW (ئەیلوولی 2026) دەهێنێتەوە: تاقیکردنەوە و هەڵسەنگاندن، لەوانەش هاوشێوەسازیی بەس نزیک لە ڕاستی.
Simulation first, then the real world, then monitoring: testing never really stops.سەرەتا ، پاشان جیهانی ڕاستەقینە، دواتر پشکنینی بەردەوام: تاقیکردنەوە هەرگیز بە تەواوی ناوەستێت.CêIT

But a simulation is only a model of the world. If it is too clean, always sunny and with sensors that never fail, the AI may pass the test and still fail outside. That is why testing continues after a system is put into use; see When AI gets it wrong.

بەڵام تەنیا وێنەیەکی سادەکراوی جیهانە. ئەگەر زۆر پاک بێت، هەمیشە خۆرەتاو و بە کە هەرگیز تێک ناچن، دەشێت تاقیکردنەوەکە تێپەڕێنێت و هێشتا لە دەرەوە سەرنەکەوێت. لەبەر ئەوە دوای ئەوەی سیستەمێک دەخرێتە کار، تاقیکردنەوە بەردەوام دەبێت؛ بڕوانە کاتێک ژیریی دەستکرد هەڵە دەکات.

Common mistakesهەڵە باوەکان

MistakeهەڵەBetterڕاستییەکەی
Logistics is just transport. تەنیا گواستنەوەیە.It is planning, supply, storage, maintenance and movement together.پلاندانان، دابینکردن، هەڵگرتن، و گواستنەوەیە پێکەوە.
A forecast is a promise.پێشبینی بەڵێنە.It is an estimate with a range; update it when things change.خەمڵاندنێکە لە نێوان کەمترین و زۆرتریندا؛ کاتێک شتەکان گۆڕان نوێی بکەرەوە.
Predictive and preventive maintenance are the same. و یەک شتن.Preventive follows a timetable; predictive follows the machine's own data. بەدوای خشتەیەکی کاتیدا دەچێت؛ بەدوای خودی ئامێرەکەدا.
The best route never changes.باشترین ڕێگا هەرگیز ناگۆڕێت.It depends on the costs; when they change, re-plan from where you are.بەندە بە کۆستەکان؛ کاتێک گۆڕان، لەو شوێنەوەی لێیت پلان دابنێرەوە.
The AI decides what is sent. بڕیار دەدات چی بنێردرێت.It suggests options; people decide and answer for the decision.هەڵبژاردە پێشنیار دەکات؛ مرۆڤ بڕیار دەدات و بەرپرسیاری بڕیارەکەیە.
A VR headset is only for games. تەنیا بۆ یارییە.Pilots, drivers, medics and rescuers train with VR and simulators.فڕۆکەوان، شۆفێر، پزیشک و فریاکەوتووان بە و ڕاهێنان دەکەن.
Passing a simulation means the system is safe.تێپەڕاندنی واتە سیستەمەکە سەلامەتە.Real-world trials and monitoring are still needed.هێشتا تاقیکردنەوەی ڕاستەقینە و پشکنینی بەردەوام پێویستن.

Exercisesڕاهێنانەکان

ExerciseڕاهێنانRe-plan by handپلاندانانەوە بە دەست

Use the map (minutes on each road). (1) Find the cheapest route from the depot S to the clinic T, and its cost. (2) The truck has just reached P when a storm report arrives: road Q–R is flooded and closed, and road R–T now takes 20 minutes. Find the cheapest route from P to T. (3) What is the total time of the trip now?

نەخشەکە بەکار بهێنە (خولەکی سەر هەر ڕێگایەک). ⁦(1)⁩ هەرزانترین ڕێگا لە کۆگای S بۆ نۆرینگەی T و کۆستەکەی بدۆزەرەوە. ⁦(2)⁩ بارهەڵگرەکە تازە گەیشتووەتە P کە ڕاپۆرتی زریانێک دێت: ڕێگای نێوان Q و R لافاوی گرتووە و داخراوە و ڕێگای نێوان R و T ئێستا ⁦20⁩ خولەک دەخایەنێت. هەرزانترین ڕێگا لە P بۆ T بدۆزەرەوە. ⁦(3)⁩ ئێستا کۆی کاتی گەشتەکە چەندە؟

Exercise map. Nodes: S (depot), P, Q, R and T (clinic). Roads with minutes: S–P 12, S–Q 18, P–Q 5, P–R 20, Q–R 9, Q–T 25, R–T 8.نەخشەی ڕاهێنان. نۆدەکان: S (کۆگا)، P، Q، R و T (نۆرینگە). ڕێگاکان بە خولەک: S بۆ P 12، S بۆ Q 18، P بۆ Q 5، P بۆ R 20، Q بۆ R 9، Q بۆ T 25، R بۆ T 8.
Find the cheapest routes before and after the storm.هەرزانترین ڕێگاکان پێش و دوای زریانەکە بدۆزەرەوە.CêIT
Show answerوەڵامەکە پیشان بدە
  • (1) S → P → Q → R → T: 12+5+9+8=3412 + 5 + 9 + 8 = 34 minutes. (S → Q → R → T takes 35, one minute longer.)
  • (2) From P: P → Q → T, 5+25=305 + 25 = 30 minutes. (P → R → T is 20+20=4020 + 20 = 40; going back through S takes 55.)
  • (3) 12+30=4212 + 30 = 42 minutes: 8 minutes more than the first plan, on safe roads.
  • ⁦(1)⁩ لە S بە P، Q و Rـدا بۆ T: 12+5+9+8=3412 + 5 + 9 + 8 = 34 خولەک. (ڕێگای S، Q، R و ⁦T 35⁩ خولەک دەخایەنێت، یەک خولەک زیاتر.)
  • ⁦(2)⁩ لە Pـەوە بە Qـدا بۆ T: 5+25=305 + 25 = 30 خولەک. (بە Rـدا 20+20=4020 + 20 = 40 خولەکە؛ گەڕانەوە بە Sـدا ⁦55⁩ خولەکە.)
  • ⁦(3)⁩ 12+30=4212 + 30 = 42 خولەک: 8 خولەک زیاتر لە پلانی یەکەم، بەسەر ڕێگای سەلامەتدا.
ExerciseڕاهێنانWhich kind of maintenance?کام جۆری چاککارییە؟

Name the approach: (1) a generator is repaired only after it stops; (2) a water pump's filters are changed every 3 months; (3) an engine is serviced when its oil sensor shows that worn-metal particles are rising.

ڕێگاکە ناو ببە: ⁦(1)⁩ مۆلیدەیەک تەنیا دوای ئەوەی وەستا چاک دەکرێتەوە؛ ⁦(2)⁩ فلتەرەکانی پەمپێکی ئاو هەموو 3 مانگ جارێک دەگۆڕدرێن؛ ⁦(3)⁩ بزوێنەرێک کاتێک ڕۆنەکەی پیشان دەدات کە وردە کانزای سواو زیاد دەبێت، چاک دەکرێتەوە.

Show answerوەڵامەکە پیشان بدە
  • (1) Reactive: fix it when it breaks.
  • (2) Preventive: a fixed timetable.
  • (3) Predictive: the machine's own sensor data says a failure is coming.
  • ⁦(1)⁩ دوای تێکچوون: کاتێک تێکچوو چاکی بکەرەوە.
  • ⁦(2)⁩ : خشتەیەکی کاتیی جێگیر.
  • ⁦(3)⁩ : خودی ئامێرەکە دەڵێت تێکچوونێک لە ڕێگادایە.
ExerciseڕاهێنانDesign an adaptive ruleڕێسایەکی خۆگونجێن دابنێ

A first-aid training app has five levels. Write simple rules for when the next task gets harder or easier, and for when the app gives a hint.

ڕاهێنانی فریاگوزاریی سەرەتایی پێنج ئاستی هەیە. ڕێسای سادە بنووسە بۆ ئەوەی کەی ئەرکی داهاتوو قورستر یان ئاسانتر دەبێت و کەی ڕێنمایی دەدات.

Hintئاماژە

Think of the ladder chart: what moves the learner up, and what moves them down?

بیر لە ئاستەکان بکەرەوە: چی فێرخوازەکە بەرز دەکاتەوە و چی دایدەبەزێنێت؟

Show answerوەڵامەکە پیشان بدە
  • After 2 successes in a row, go up one level (never above 5).
  • After a mistake, stay or go down one level (never below 1) and show a hint on the step that went wrong.
  • If the same skill fails twice, give an extra practice task on that skill before moving on.
  • Keep a record of each skill, so the next lesson starts where the learner is weak.
  • دوای دوو سەرکەوتن لە دوای یەک، یەک ئاست بەرز ببەرەوە (هەرگیز سەرووی 5 نا).
  • دوای هەڵەیەک، لە هەمان ئاست بمێنەرەوە یان یەک ئاست دابەزە (هەرگیز خوار 1 نا) و ڕێنمایی لەسەر ئەو هەنگاوەی هەڵە بوو پیشان بدە.
  • ئەگەر هەمان کارامەیی دوو جار سەرنەکەوت، پێش ئەوەی بچیتە پێشەوە ئەرکێکی ڕاهێنانی زیادە لەسەر ئەو کارامەییە بدە.
  • تۆماری هەر کارامەییەک بهێڵەرەوە، بۆ ئەوەی وانەی داهاتوو لەو شوێنەوە دەست پێ بکات کە فێرخوازەکە تێیدا لاوازە.

Summaryپوختە

  • Logistics gets the right supplies to the right place at the right time along a supply chain; AI forecasts needs, predicts breakdowns and plans routes, while people decide.
  • Predictive maintenance: sensors → model → 'likely to fail soon' → a planned repair. Preventive maintenance follows a timetable; reactive maintenance waits for the breakdown.
  • Route optimisation is a search on a weighted graph; when a storm changes the costs, re-plan from where the truck is now. The result is an expected benefit, not a promise.
  • The same tools serve humanitarian work: WFP's Optimus suggests food baskets, sourcing and delivery plans, and people decide.
  • Simulation and VR let people practise dangerous or rare situations safely; adaptive training fits each learner; an AI opponent gets harder as the trainee improves.
  • Simulation also tests AI systems before use, but it must be realistic, and real-world testing and monitoring must follow.
  • پێداویستیی گونجاو بە درێژایی لە کاتی گونجاودا دەگەیەنێتە شوێنی گونجاو؛ پێشبینیی پێداویستییەکان و تێکچوونەکان دەکات و ڕێگا پلان دەکات، لە کاتێکدا مرۆڤ بڕیار دەدات.
  • : ← ← «بەم زووانە تێک دەچێت» ← چاککردنەوەی پلانکراو. بەدوای خشتەیەکدا دەچێت؛ دوای تێکچوون چاوەڕێی تێکچوونەکە دەکات.
  • گەڕانە لە ؛ کاتێک زریانێک کۆستەکان دەگۆڕێت، لەو شوێنەوەی بارهەڵگرەکە ئێستا لێیەتی پلان دابنێرەوە. ئەنجامەکە سوودێکی چاوەڕوانکراوە، نەک بەڵێن.
  • هەمان ئامرازەکان خزمەتی کاری مرۆیی دەکەن: Optimusـی WFP سەبەتەی خۆراک، شوێنی کڕین و پلانی گەیاندن پێشنیار دەکات و مرۆڤ بڕیار دەدات.
  • و ڕێگە دەدەن خەڵک بە سەلامەتی لەسەر بارودۆخی مەترسیدار یان دەگمەن ڕاهێنان بکەن؛ لەگەڵ هەر فێرخوازێک دەگونجێت؛ ڕکابەرێکی بە باشتربوونی ڕاهێنراوەکە قورستر دەبێت.
  • سیستەمەکانی پێش بەکارهێنان تاقی دەکاتەوە، بەڵام دەبێت نزیک لە ڕاستی بێت و دەبێت تاقیکردنەوەی ڕاستەقینە و پشکنینی بەردەوام بەدوایدا بێن.
Review questionsپرسیارەکانی پێداچوونەوە
What is the difference between preventive and predictive maintenance?جیاوازیی نێوان چاککاریی پێشوەختە و چاککاریی پێشبینیکەر چییە؟

Preventive maintenance follows a fixed timetable (for example every 250 hours). Predictive maintenance repairs a machine when its own sensor data shows that a failure is coming.

بەدوای خشتەیەکی کاتیی جێگیردا دەچێت (بۆ نموونە هەر ⁦250⁩ کاتژمێرێک). ئامێرێک کاتێک چاک دەکاتەوە کە خۆی پیشان دەدات تێکچوونێک لە ڕێگادایە.

Why must the planner re-plan from the truck's current position?بۆچی دەبێت پلاندانەر لە شوێنی ئێستای بارهەڵگرەکەوە پلاندانانەوە بکات؟

The truck has already driven part of the route; the best path now is the cheapest one from where it is, with the new costs.

بارهەڵگرەکە بەشێک لە ڕێگاکەی بڕیوە؛ باشترین ڕێگا ئێستا هەرزانترینیانە لەو شوێنەوەی لێیەتی، بە کۆستە نوێیەکان.

How does an AI opponent in a simulation get harder?ڕکابەرێکی ژیریی دەستکرد لە هاوشێوەسازییەکدا چۆن قورستر دەبێت؟

It uses smarter tactics each round, for example by searching more moves ahead (as with the minimax depth) or by learning from many games.

لە هەر گەڕێکدا تاکتیکی زیرەکتر بەکار دەهێنێت، بۆ نموونە بە گەڕان بە جووڵەی زیاتر بەرەو پێشەوە (وەک قووڵایی ) یان بە فێربوون لە زۆر یاری.

Lesson complete!وانەکە تەواو بوو!

Well done. Review the key terms, then go on to the next lesson.دەستخۆش. سەیرێکی زاراوە سەرەکییەکان بکەرەوە، پاشان بڕۆ بۆ وانەی داهاتوو.

Key termsزاراوە سەرەکییەکان

Next lessonوانەی داهاتوو

Decisions, Human Control and Responsibilityبڕیار، کۆنترۆڵی مرۆڤ و بەرپرسیارێتی

References and further readingسەرچاوەکان و خوێندنەوەی زیاتر

This lecture is a first edition. The Kurdish text and terminology are waiting for review by Kurdish IT teachers.ئەم وانەیە چاپی یەکەمە. دەقە کوردییەکە و زاراوەکانی چاوەڕێی پێداچوونەوەی مامۆستایانی کوردی بواری IT ن.

Appearanceڕووکار

Modeدۆخ
Styleشێواز
Coloursڕەنگەکان

Saved on this device only.تەنیا لەسەر ئەم ئامێرە پاشەکەوت دەکرێت.