Intermediate · story · Level 3 · 5:20
Intermediate Spanish: ANYmal Learns to Walk
Watch this episode
This factual lesson explains how researchers trained the ANYmal quadruped robot to develop locomotion through reinforcement learning. The robot practiced movements in simulation, received feedback, and transferred its learned control policy to the real world, where it handled uneven ground and disturbances.
Story context, sources & AI disclosure
This factual lesson explains how researchers trained the ANYmal quadruped robot to develop locomotion through reinforcement learning. The robot practiced movements in simulation, received feedback, and transferred its learned control policy to the real world, where it handled uneven ground and disturbances. Create your own personalized learning podcast Want a podcast built around your interests and learning needs? Sign up for LingoTailor and create a personalized podcast that evolves as you learn. Register at https://lingotailor.clipiti.com This episode uses AI-generated content and audio and was published automatically, without human review. Language explanations and factual content may contain mistakes.
Account required. A limited lifetime trial is available; your allowance is shown before generation.
Transcript & learning notes
Follow along, pause, and replay the Spanish passages.
EnglishWelcome to LingoTailor, where you learn Spanish through stories. First, you'll hear the story in English to understand the context. Then you'll hear it in Spanish, learn useful vocabulary, and listen to the Spanish story again. Today's story is: ANYmal Learns to Walk.
EnglishHere is the English version.
EnglishIn 2019, researchers led by Hwangbo trained ANYmal, a four-legged robot, to walk. Its goal was simple, but the method was not. Instead of programming every footstep, the team trained a control policy with reinforcement learning. In simulation, ANYmal tried movements and received feedback about its motion and stability. Many attempts failed, so the system gradually changed the policy and favored better movements. The main obstacle was the gap between simulation and the real world: a virtual robot does not experience exactly the same surfaces, delays, or physical forces. The researchers transferred the learned policy to the real ANYmal and tested it on uneven ground and under external disturbances. The robot could recover its balance and continue walking. This documented experiment showed that trial and error, when guided by feedback, can produce a flexible walking skill without specifying every step in advance.
EnglishNow hear the same story in Spanish.
EspañolAquí está la versión en español.
EspañolEn 2019, investigadores dirigidos por Hwangbo entrenaron a ANYmal, un robot de cuatro patas, para caminar. Su objetivo era sencillo, pero el método no lo era. En vez de programar cada paso, el equipo entrenó una política de control mediante aprendizaje por refuerzo. En una simulación, ANYmal probaba movimientos y recibía información sobre su desplazamiento y su estabilidad. Muchos intentos fallaban, así que el sistema modificaba poco a poco la política y favorecía los movimientos más eficaces. El principal obstáculo era la diferencia entre la simulación y el mundo real: un robot virtual no experimenta exactamente las mismas superficies, demoras ni fuerzas físicas. Para superar un obstáculo tan importante, los investigadores transfirieron la política aprendida al ANYmal real y lo probaron sobre terreno irregular y frente a perturbaciones externas. El robot podía recuperar el equilibrio y seguir caminando. A base de ensayo y error, logró dar con la solución sin que los investigadores especificaran cada paso de antemano: partir de cero no significó avanzar sin guía, sino aprender mediante la información de cada intento.
EnglishHere are four reusable expressions from the Spanish version.
Española base de ensayo y error
EnglishBy repeatedly trying, receiving results, and changing the approach. It emphasizes learning through attempts rather than through a complete plan from the beginning.
EspañolEl equipo mejoró el diseño a base de ensayo y error.
Españolsuperar un obstáculo
EnglishTo overcome a difficulty that blocks progress, either literally or figuratively.
EspañolTuvieron que superar un obstáculo técnico antes de continuar.
Españoldar con la solución
EnglishTo find the answer or method that solves a problem, often after searching or testing several possibilities.
EspañolDespués de varias pruebas, dieron con la solución.
Españolpartir de cero
EnglishTo begin without an existing result, system, or prepared foundation.
EspañolEl equipo tuvo que partir de cero para crear el prototipo.
EnglishListen once more to the Spanish version.
EspañolEn 2019, investigadores dirigidos por Hwangbo entrenaron a ANYmal, un robot de cuatro patas, para caminar. Su objetivo era sencillo, pero el método no lo era. En vez de programar cada paso, el equipo entrenó una política de control mediante aprendizaje por refuerzo. En una simulación, ANYmal probaba movimientos y recibía información sobre su desplazamiento y su estabilidad. Muchos intentos fallaban, así que el sistema modificaba poco a poco la política y favorecía los movimientos más eficaces. El principal obstáculo era la diferencia entre la simulación y el mundo real: un robot virtual no experimenta exactamente las mismas superficies, demoras ni fuerzas físicas. Para superar un obstáculo tan importante, los investigadores transfirieron la política aprendida al ANYmal real y lo probaron sobre terreno irregular y frente a perturbaciones externas. El robot podía recuperar el equilibrio y seguir caminando. A base de ensayo y error, logró dar con la solución sin que los investigadores especificaran cada paso de antemano: partir de cero no significó avanzar sin guía, sino aprender mediante la información de cada intento.
EnglishMake your learning personal with LingoTailor: podcasts adapted to your interests, level, and progress. Practice speaking with ChatGPT, get feedback, and shape your next episode. Follow the link in the description to get started. Subscribe now so you never miss a new episode. Don't just listen to a new language. Start speaking it.