HejuruUbuyobozi bukurikira
Inzira no Kurangiza-Impera yo gutwara
Ibigo
AMABWIRIZA
Wayve’s LINGO research explores how language can support interpretation and instruction of learned driving systems.
Umva iki gitaboAI-byavuzwe · ~2 min
LINGO-1 studied natural-language commentary for driving; LINGO-2 connects vision, language, and driving action in a closed-loop model. Explanations can make behavior easier to inspect, but do not by themselves prove that a model’s stated reason faithfully captures its internal causal process.
Wayve is a London-based autonomous-driving company developing learned driving systems. LINGO-1 explored open-loop natural-language commentary and question answering about driving scenes. LINGO-2 extended the work to a closed-loop vision-language-action model that takes images, driving data, and language, and can produce both driving behavior and text. Wayve reports testing the model on public roads; this research result should not be read as proof of general autonomous-driving safety. LingoQA is a separate video question-answering dataset and benchmark for evaluating vision-language models on autonomous-driving questions.
Reviewing language commentary associated with a driving model’s behavior during research and testing
Using constrained navigation commands to study how language can condition driving behavior
Asking video question-answering benchmarks about objects, actions, and reasons in driving scenes
Pairing driving video with language annotations to evaluate vision-language systems
Research on language and driving continues to study instruction following, scene questions, and links between explanations and action. Useful evaluations should separate task performance from explanation faithfulness and road safety. Wayve describes an AI Driver foundation-model strategy for working with automakers; deployment still depends on validation, local requirements, and the capabilities of the specific system.
Ibaruwa y'amakuru y'ubuntu
Inkuru eshatu zizewe za AI buri gitondo cy'iminsi y'akazi, zanditse mu rurimi rworoshye. Ubuntu iteka ryose, nta kwamamaza.
Imeri imwe buri munsi y'akazi. Kwiyimura mu buryo bumwe. Ntitwigera tugurisha cyangwa ngo dusangire aderesi yawe.
Isuzume
Ibisubizo ako kanya kuri buri kibazo, n'icyemezo gishobora gusangirwa gifite ID yagenzurwa umaze gutsinda isomo.
Fasha uburezi bwa AI bw'ubuntu. AI Understanding ni umuryango utegamiye kuri leta wa 501(c)(3) — nta kwamamaza, nta kwishyura, iteka ryose. Tanga impano
Wayve’s LINGO research explores how language can support interpretation and instruction of learned driving systems. LINGO-1 studied natural-language commentary for driving; LINGO-2 connects vision, language, and driving action in a closed-loop model. Explanations can make behavior easier to inspect, but do not by themselves prove that a model’s stated reason faithfully captures its internal causal process.
LINGO-1 yerekanye icyerekezo-cyerekezo cyerekana imvugo yo gufata ibyemezo mucyongereza cyoroshye, bigatuma sisitemu isobanurwa neza.
LINGO-2 nicyitegererezo-ururimi-ibikorwa-byerekana: ihuza inyandiko na politiki nyayo yo gutwara, bityo irashobora kuyoborwa namabwiriza no gusobanura imyitwarire yayo.
Wayve yiga imyitwarire yo gutwara biturutse kumibare hamwe numuyoboro uhuza imitsi, aho kwishingikiriza kumategeko yagutse yakozwe namaboko namakarita arambuye.
Wayve says the same model generates driving behavior and textual predictions; this ties language to model outputs but does not prove that every explanation faithfully captures internal causes.
Wayve igamije gutanga icyitegererezo cyayo cyo gutwara ibinyabiziga kubakora ibinyabiziga aho gukora ibinyabiziga byayo.
Komeza wige
Abandi bayobozi batoranijwe kuriyi ngingo
HejuruUbuyobozi bukurikira
Inzira no Kurangiza-Impera yo gutwara
Ibigo