Unsupervised Learning for Physical Interaction through Video Prediction (Chelsea Finn, Ian Goodfellow & Sergey Levine)

Sandeep Prasada argues that commonsense concepts have a formal structure that guides how we think, speak and reason.Conceptual Form distinguishes what we know about a category from the perspectives through which we understand its members.The book develops an instance-of-kind perspective and examines how people reason about kinds, individuals and their properties.It applies this account to generic statements, systematic polysemy and metaphor, connecting conceptual structure with language.The later chapters consider how these perspectives are acquired and what they imply for theories of conceptual knowledge.Author: Sandeep PrasadaSource: Conceptual Form: The Hidden Structure of Commonsense Concepts · Sandeep PrasadaSeries: Mind & Psychology · Book deep diveExplore: Mind & PsychologyRelated episodes: How Children Learn the Meanings of Words (Paul Bloom) · Thought and Language (Lev Vygotsky)Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

Chelsea Finn, Ian Goodfellow and Sergey Levine investigate how robots can learn the effects of physical interactions from unlabeled video.Their action-conditioned models predict how pixels move between frames, reusing visual appearance rather than reconstructing every future image from scratch.The paper introduces a dataset of 59,000 robot-pushing interactions and tests predictions on previously unseen objects.Its experiments compare alternative motion-prediction architectures and examine their ability to produce plausible futures conditioned on a robot's actions.This 2016 work provides an early foundation for robotic world models and connects visual prediction with the longer-term goal of planning through learned models.Author: Chelsea Finn, Ian Goodfellow and Sergey LevineSource: Unsupervised Learning for Physical Interaction through Video Prediction · Chelsea Finn, Ian Goodfellow and Sergey LevineSeries: Weekly AI Papers · Historical classicExplore: AI & Computer Science · AI PapersRelated episodes: Backpropagation: Learning Representations by Back-Propagating Errors (Rumelhart, Hinton & Williams) · Robotics, Vision and Control (Peter Corke) — Part 3 of 5: Robot ManipulatorsGot something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

Mateo Clémente and colleagues introduce La-JEPA, a compact robot policy trained in the feature space of a pretrained V-JEPA 2.1 video encoder.The model learns to predict future visual features alongside actions during training, then generates actions alone when controlling the robot.The authors evaluate it on LIBERO-Plus, RoboTwin 2.0 and four real-robot tasks, reporting strong results with smaller policies than several generation-based alternatives.Their design study compares JEPA features with video-VAE latents and examines which forms of predictive supervision improve control.The paper also identifies corrupted instructions in the released LIBERO-Plus evaluator and re-evaluates baselines under a corrected protocol.Author: Mateo Clémente, George Z. Xue, Amin Abyaneh, Liheng Ma, Rui Heng Yang, Tongtong Cao, Zhanguang Zhang and Yingxue ZhangSource: La-JEPA: Compact Robot Policies with Latent World Modeling · Mateo Clémente, George Z. Xue, Amin Abyaneh, Liheng Ma, Rui Heng Yang, Tongtong Cao, Zhanguang Zhang and Yingxue ZhangSeries: Weekly AI Papers · Recent researchExplore: AI & Computer Science · AI PapersRelated episodes: Yann LeCun & Randall Balestriero: The Platonic Representation Hypothesis with JEPAs · Robotics, Vision and Control (Peter Corke) — Part 5 of 5: Vision-Based ControlGot something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

The final part connects the series’ robotics and vision foundations through vision-based control.Position-based visual servoing uses estimated spatial quantities, while image-based visual servoing acts on errors measured in image features.The interaction matrix links motion to changes in the image, making the choice of features, depth estimates and camera model consequential.Advanced methods include partitioned, polar-coordinate and spherical-camera approaches to visual servoing.Applications to robot arms, mobile robots and aerial robots close the perception-to-action loop and expose the assumptions and performance limits of the methods.Author: Peter CorkeSource: Robotics, Vision and Control: Fundamental Algorithms in Python, third edition (2023) · Peter CorkeSeries: Robotics, Vision and Control — Part 5 of 5 · Vision-Based ControlExplore: AI & Computer Science · AI BooksRelated episodes: How the Body Shapes the Way We Think: A New View of Intelligence (Pfeifer & Bongard) · Structure and Interpretation of Computer Programs (Harold Abelson & Gerald Jay Sussman) · Lie Groups, Lie Algebras, and Representations (Brian Hall)Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

Part 4 follows the book’s five computer-vision chapters, from the physical properties of light to information recovered from multiple images.Light, color and image-processing operations explain how measurements become usable digital representations.Region, line and point features provide ways to describe structures and establish correspondences in images.Camera models and calibration relate image measurements to the geometry of the world.Multiple-view geometry and stereo methods show how relationships between images reveal depth, structure and motion, preparing the transition to visual feedback control.Author: Peter CorkeSource: Robotics, Vision and Control: Fundamental Algorithms in Python, third edition (2023) · Peter CorkeSeries: Robotics, Vision and Control — Part 4 of 5 · Computer VisionExplore: AI & Computer Science · AI BooksRelated episodes: How the Body Shapes the Way We Think: A New View of Intelligence (Pfeifer & Bongard) · Structure and Interpretation of Computer Programs (Harold Abelson & Gerald Jay Sussman) · Lie Groups, Lie Algebras, and Representations (Brian Hall)Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

Part 3 turns from moving a robot’s base to controlling the motion of a robot arm.Forward and inverse kinematics connect joint configurations with the position and orientation of the end effector.The manipulator Jacobian relates joint velocity to task-space motion and reveals the significance of singularities and manipulability.Dynamics adds mass, inertia and the forces or torques required to produce a desired movement.Corke’s control methods connect these models with the practical problem of making an arm follow a commanded motion.Author: Peter CorkeSource: Robotics, Vision and Control: Fundamental Algorithms in Python, third edition (2023) · Peter CorkeSeries: Robotics, Vision and Control — Part 3 of 5 · Robot ManipulatorsExplore: AI & Computer Science · AI BooksRelated episodes: How the Body Shapes the Way We Think: A New View of Intelligence (Pfeifer & Bongard) · Structure and Interpretation of Computer Programs (Harold Abelson & Gerald Jay Sussman) · Lie Groups, Lie Algebras, and Representations (Brian Hall)Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

Part 2 follows Corke’s treatment of mobile robots, navigation, localization and mapping.Vehicle models explain how wheeled and aerial robots can move, and which motions their physical constraints permit.Reactive behavior and map-based path planning offer different ways to reach a destination while handling obstacles.Odometry and probabilistic localization address the uncertainty in a robot’s estimate of where it is.Mapping and SLAM bring motion, measurements and environmental structure together, showing why navigation depends on both geometry and reliable estimation.Author: Peter CorkeSource: Robotics, Vision and Control: Fundamental Algorithms in Python, third edition (2023) · Peter CorkeSeries: Robotics, Vision and Control — Part 2 of 5 · Mobile RoboticsExplore: AI & Computer Science · AI BooksRelated episodes: How the Body Shapes the Way We Think: A New View of Intelligence (Pfeifer & Bongard) · Structure and Interpretation of Computer Programs (Harold Abelson & Gerald Jay Sussman) · Lie Groups, Lie Algebras, and Representations (Brian Hall)Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

The first part introduces Peter Corke’s Robotics, Vision and Control and the questions that connect sensing, geometry and action.Chapter 1 places mobile robots, robot arms and robotic vision in context, including the ethical responsibilities of building autonomous systems.Chapter 2 develops coordinate frames, relative pose and the representations used to describe position and orientation in two and three dimensions.Chapter 3 connects those geometric descriptions with velocity, acceleration, forces and smooth trajectories through time.Inertial sensing and sensor fusion show how these foundations support a robot’s estimates of its own motion, preparing the ground for mobile robotics.Author: Peter CorkeSource: Robotics, Vision and Control: Fundamental Algorithms in Python, third edition (2023) · Peter CorkeSeries: Robotics, Vision and Control — Part 1 of 5 · Introduction and FoundationsExplore: AI & Computer Science · AI BooksRelated episodes: How the Body Shapes the Way We Think: A New View of Intelligence (Pfeifer & Bongard) · Structure and Interpretation of Computer Programs (Harold Abelson & Gerald Jay Sussman) · Lie Groups, Lie Algebras, and Representations (Brian Hall)Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

Pamela S. Nadell traces the persistence of antisemitism in America from the arrival of Jews in New Amsterdam in 1654 to contemporary attacks.The book follows religious hostility, political discrimination and social exclusion through successive periods of American history.It examines barriers to citizenship, education and employment alongside propaganda, conspiracy thinking and violence.Nadell also follows Jewish responses to prejudice and the tension between lived discrimination and America’s promise of religious liberty.Her history asks how antisemitism has endured and adapted within a country often understood as a refuge from persecution.Author: Pamela S. NadellSource: Antisemitism, an American Tradition · Pamela S. NadellSeries: History & JudaismExplore: History & JudaismRelated episodes: The Final Solution to the Jewish Question: A Love Story for the Living (Dara Horn) · Jew-Hatred: The Latest Edition (Simon Schama) · Confronting Antisemitism: Fighting Antisemitism Today (Theodor Adorno)Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

Leovino Madriaga Garcia reads Paul Ricoeur’s early philosophy through the relationship between responsibility and hope.The book follows Freedom and Nature, Fallible Man and The Symbolism of Evil to examine willing, human limitation and the acknowledgment of fault.Garcia explains how phenomenology and the interpretation of symbols offer different approaches to understanding human action.He then connects these concerns with Ricoeur’s early essays on civilization, Christian social responsibility, politics and the state.The central question is how an honest account of human fallibility can sustain responsible action and hope for change.Author: Leovino Madriaga GarciaSource: Ricoeur’s Early Ethical Philosophy: Explorations of Responsibility and Hope · Leovino Madriaga GarciaSeries: Philosophy & Critical TheoryExplore: Philosophy & Critical TheoryRelated episodes: Memory, History, Forgetting (Paul Ricoeur): Memory and Historical Truth · Time and Narrative (Paul Ricoeur): How Stories Shape TimeGot something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!